Spatial patterns associated with the distribution of immature stages of Aedes aegypti in three dengue high-risk municipalities of Southwestern Colombia

Aedes aegypti mosquitoes are the main vector of human arbovirosis in tropical and subtropical areas. Their adaptation to urban and rural environments generates infestations inside households. Therefore, entomological surveillance associated with spatio-temporal analysis is an innovative approach for vector control and dengue management. Here, our main aim was to inspect immature pupal stages in households belonging to municipalities at high risk of dengue in Cauca, Colombia, by implementing entomological indices and relating how they influence adult mosquitos’ density. We provide novel data for the geographical distribution of 3,806 immature pupal stages of Ae. aegypti. We also report entomological indices and spatial characterization. Our results suggest that, for Ae. aegypti species, pupal productivity generates high densities of adult mosquitos in neighbouring households, evidencing seasonal behaviour. Our dataset is essential as it provides an innovative strategy for mitigating vector-borne diseases using vector spatial patterns. It also delineates the association between these vector spatial patterns, entomological indicators, and breeding sites in high-risk neighbourhoods.


Introduction
The Aedes aegypti mosquito is the main vector of human arbovirosis, encompassing diseases such as yellow fever, dengue, chikungunya, and Zika, which are prevalent in tropical and subtropical areas [1].Its adaptation to local anthropic conditions, urban environments, together with the optimal temperature, precipitation, and relative humidity, among other environmental and socioeconomic factors, has given this mosquito the ability to establish reproductive populations spanning both urban and rural domains [2][3][4].In Colombia, Ae. aegypti is the main vector of the dengue virus and is reported at altitudes between sea level and 2,302 m.Therefore, the significant presence and abundance of this vector threaten public health within the country [5,6].
Direct proximity to humans has allowed the hematophagous female to feed on human or domestic animal blood, causing infestations in different types of containers in and around households.In this context, the presence of artificial and natural containers, such as tanks, drums, bottles, tires, aquatic plants (bromeliads and water retained in plant axils), stagnant water, and pet drinking troughs, serve as potential breeding grounds for the immature stages of Ae. aegypti [7,8].The surveillance of these types of containers helps reduce the proliferation of pupae and, therefore, mosquito densities.Hence, mosquito sampling is an indispensable tool for designing approaches of entomological surveillance of the vector specific to each situation [9].
Following the methodologies established for Colombia [7,10], the inspection of potential breeding sites is determined with entomological indicators that make it possible to identify the index of pupae per person, pupal productivity, and the presence of immature stages in dwellings [7,10].However, this is not done in all regions.The importance of using this type of entomological index lies in providing a representative approximation of the local adult populations of mosquitoes while requiring a relatively small sample size for household inspection [9,11].Furthermore, the index of pupae per person makes it possible to calculate a minimum level of Ae. aegypti pupae infestation that significantly enhances the risk of dengue virus (DENV) transmission.Specifically, if the index of pupae per person ranges from 0.5 to 1.5 at 28 °C with 0% to 67% seroprevalence [12], making it possible to identify areas at higher risk.Surveillance of Ae. aegypti mosquitoes has mainly relied on qualitative larval indices, bearing little relation to the number of female mosquitoes.Moreover, Ae. aegypti populations have been reported to be spatially heterogeneous in some areas [11].Thus, using spatio-temporal analyses to predict Ae. aegypti hot spots, i.e., groups of dwellings with higher pupal productivity, high vector density, and potential DENV transmission, is an innovative strategy that could help the prevention, vector control, and integrated management of dengue in high-risk municipalities.This is because this method allows detecting patterns of geographical distribution of the vector [11,13].Consequently, it is fundamental to embrace this kind of infestation indices because immature stages, specifically the pupae, represent a precursor stage to adult mosquitos with very low mortality.This increases the correlation between Ae. aegypti density and dengue virus [7,14].
Thus, in this study, entomological surveillance was compared using quantitative indices of pupae and adults in three municipalities at high dengue risk: Patía (El Bordo), Miranda, and Piamonte in the Cauca department.Specifically, our study aimed at examining the influence of pupal productivity, among other entomological indices, on the density of adult mosquitoes, and their spatial and temporal patterns.Spatial regression and bivariate spatial autocorrelation analyses were carried out, allowing to model and explore spatial relationships, while capturing the relationship between variables even when they did not overlap within the same space.The immature stages (pupae) were collected from the inspection of containers smaller or larger than 20 l, including tanks, drums, and tires, among others, from inside the houses of three municipalities.The municipalities were selected as study areas based on their endemic-epidemic behaviours for DENV transmission, as they are characterized by focal endemics, heterogeneous transmission scenarios, and temporal and cyclical patterns of at-risk populations.
The resulting pupae dataset is in Darwin Core file format, with 73 terms available.We included all mandatory fields, which were submitted to the Integrated Publishing Toolkit (IPT) for review by SiB-Colombia.Metadata fields are also available from the IPT website [15].A total of 3,480 adult individuals of the family Culicidae are included in this dataset.Of these, about 69% of the specimens were previously reported in 2021 [16], while 1,078 are the new records collected in 2022.
These biological registers are fundamental for the scientific community as they provide the geographical coordinates and entomological indices necessary for the public health surveillance performed by health entities.In particular, these measures are essential for preventing and controlling the etiological agent from houses and neighbourhoods with high infestation rates.

Sampling
The specimens of the immature stages of the Ae.aegypti pupae belonged to dwellings located in the municipalities of Patía (El Bordo), Miranda, and Piamonte, within the department of Cauca, Colombia (Figure 1).These municipalities were selected because they were considered endemic-epidemic with a high dengue transmission risk after conducting a spatiotemporal analysis [16].The sample size was delimited spatial scale of blocks by Kernel density analysis, georeferencing the dengue cases reported from 2015 to 2019 in the urban areas of each municipality.Additionally, the sample size was calculated from the estimated prevalence of dengue (10.5%) in the municipality [17] with a confidence level of 99% calculated by the Epi Info™ software, using the estimated population size and the clusters obtained in the Kernel analysis.

Species collection
A survey of the dwellings was carried out during the day between 8:00 a.m. and 5:00 p.m.
Then, the parameters stipulated in the guide Management for the entomological surveillance and control of dengue transmission were followed [7,10].Artificial water containers found in the households were selected.All breeding sites were inspected regardless of their size, i.e., containers smaller than 20 l, such as bottles, vases, cans, tires, and small plastic containers.The pupae were collected with a Pasteur pipette and then counted.For containers larger than 20 l, the number of pupae collected with a sweep net was counted, and the volume of the tanks for water storage (TWS) and the existing water level were evaluated to calculate the calibration factor [7] and obtain the estimated number of pupae or the pupal productivity.
Entomological inspections for adult mosquitoes were conducted during specific months in 2021 (February, March, April, July, October, and November) and 2022 (February, March, May, and July).These inspections were performed between 8:00 a.m. and 5:00 p.m., averaging 10 min per house.In each dwelling, a search for adult mosquitoes was conducted in the living room, dining room, bathrooms, kitchens, laundry yard, and other areas; the mosquitoes were captured using a Prokopack aspirator.Using these procedures, during the examination, special attention was directed toward exploring shaded areas and locations near water containers.Calculate the estimated number of pupae per container.For deposits smaller than 20 l, only the number of pupae is counted.For deposits larger than 20 l, multiply the number of pupae by calibration factor (c.f.) provided by water level.Water levels: <1/3 c.f. not applicable; 1/3 c.f.:

Species classification and spatial characterization
After the collection of the immature stages (pupae) and adults, the species were identified and taxonomically classified by entomological experts supported by the taxonomic keys developed by Forattini [19] (1995) and Harrison et al. [20] (2016) to differentiate them from the immature stages and adults belonging to other species.Continuing with the field protocol, 3,806 pupae of Ae. aegypti were identified, of which 395 were preserved in 0. The entomological information collected was recorded using the ArcGIS ® Survey 123 application [22], which provided the geographic location of each specimen collected.A code was associated to each vial with the socio-demographic information of the survey.The total number of pupae recorded in each survey was used to determine the entomological indices of pupal productivity, female productivity, pupal index per person, Breteau's pupal index [7,10], and Aedes sp.pupae sex ratio F:M (Table 1).
The entomological information, together with the geographical block information of the municipalities, was used to perform scatterplot analyses (R 2 , slope b, p-value), as well as scatterplot matrices to evaluate the relationships between entomological variables: number of adult Ae. aegypti mosquitoes [18], total number of pupae, pupal productivity, female productivity, and number of pupae per person vs. the frequency of tanks for water storage, miscellaneous containers smaller than 20 l, and drums.
A global bivariate Moran Index analysis (p ≤ 0.  The above analyses were performed for two sampling periods in each municipality.For the municipality of Patía (El Bordo), samples were collected in March and October 2021, for Miranda samples were collected in April and November 2021, and for Piamonte samples were collected in July 2021 and February 2022, since the rainfall pattern was seasonal between these months.Finally, the analyses were performed using the GeoDa v1.20 program [23] and the maps were visualized using the ArcGIS ® 10.8 software (RRID:SCR_011081).

DATA VALIDATION AND QUALITY CONTROL
A total of 3,480 adult specimens were identified across 1,200 households, comprising 1,459 females and 2,021 males.In 2021, 2,402 records were documented, while 1,078 new records were added in 2022.In terms of municipal distribution, Miranda reported 500 individuals (14.36%),Patía 1,305 (37.5%), and Piamonte had the highest count with 1,675 (48.13%) adult individuals.Notably, 71 individuals were found within a Patía household, and an additional 125 individuals were identified within a different household in Piamonte.
Moreover, a total of 3,806 immature specimens (pupae) of Ae. aegypti were identified, distributed among 160 records (positive dwellings out of the total inspected).For the municipality of Patía (El Bordo), 1,493 individuals were found in 67 positive dwellings, for Miranda 1,173 individuals were in 58 dwellings, and for Piamonte 1,140 individuals were found in 35 dwellings.The neighbourhoods with the highest number of specimens were Villa Los Prados (n = 519) and La Paz (n = 359) in Piamonte, followed by the San Antonio neighbourhood (n = 391) in Miranda, and the Libertador (n = 441) and Olaya Herrera (n = 438) neighbourhoods in Patía.It should be noted that 70 of the inspected dwellings were found with more than 15 pupae inside the house.The largest number of detected specimens belonged to a house in the municipality of Piamonte, with 280 pupae (Table 2).
For the wet season of the sampling, our spatial regression analysis found a positive correlation for the municipalities.Specifically, the determination coefficient between the captures of pupae and adults was higher than 50% in the municipalities of Patía (R = 0.7292; R 2 = 0.53) and Miranda (R = 0.7486; R 2 = 0.56) and lower in Piamonte (R = 0.4252; R 2 = 0.18), highlighting the relevance of the adult index for entomological surveillance.In addition, a spatial autocorrelation was observed between the presence of pupa-positive houses and a higher density of adults in neighbouring blocks.
For the Patía municipality, the variable productivity of female pupae explained the number of adult mosquitoes with a model adjustment of 53.2% in the period of higher rainfall and with an adjustment of 8.4% in the period of lower rainfall, presenting in both cases positive autocorrelation between the variables (Table 3).This allowed us to locate the clusters in which pupal productivity generated a high mosquito density at the block level (Figure 2).
Variables such as the frequency of positive vases when rainfall reached only 1 mm (October 2021) explained the female pupal productivity by only 15%.However, in the period of higher precipitation (211 mm, March 2021), the tank container explained 37% of the female pupal productivity and, subsequently, the density of adult mosquitoes.This comparison is interesting, considering that the climate of the Patía municipality is bimodal, with peaks in April and November [26].Our local analysis (LISA) observed heterogeneous high-high and low-low spatial clusterings for each season, mainly in the Olaya Herrera neighborhood (March and October 2021).
The model for the municipality of Miranda showed an R 2 model fit of 56% and a positive spatial autocorrelation in the season of lower rainfall, although with a high percentage of relative humidity, among the variables of female pupal productivity, explaining the number of adult mosquitoes.Positive autocorrelation was also found for the variables frequency-of-positive-low-TWS and pupae-productivity.However, for the period of higher precipitation and lower relative humidity, a negative spatial autocorrelation was found for the variables evaluated, i.e., with clustering patterns close to randomness (Table 3).
For Piamonte, during the wet season, the female-pupal-productivity variable explained the model fit (R 2 ) by 18%, with a positive bivariate spatial autocorrelation (Moran Index (MI) = 0.058).Miscellaneous containers smaller than 20 l, drums, and TWS presented non-significant positive autocorrelation of entomological indicators and number of adults.This latter finding is in contrast with the correlation between the percentage of positive buckets and pupae productivity (R 2 = 37.7%) in the dry season of sampling.These findings suggest that for the species Ae. aegypti, pupal productivity generates high densities of adults in neighbouring houses, allowing us to identify with the local indicator of spatial association (LISA) the blocks of neighbourhoods where this trend occurs, evidencing seasonal behaviour.While the spatial correlation linked the capture of pupae and adults in the same geographical space, the bivariate autocorrelation (MI) related these same variables, without necessarily coinciding in the same dwelling.Likewise, between municipalities, a greater positive spatial autocorrelation between pupae and adults was observed in Patía, especially when precipitation decreased before the onset of the rainy season.Our results are similar to other spatiotemporal studies of mosquito density [13,27], showing heterogeneous patterns of occurrence for each territory, with seasonal behaviours registering higher infestation rates in seasons of higher rainfall.

RE-USE POTENTIAL
Our database and vector distribution map provide important resources for understanding the spatial patterns of the vector and its relationship with entomological indicators and breeding sites, which could increase dengue virus transmission in the municipalities.
To improve the accessibility and usability of these data, they have been included in the GBIF.These data will be useful for making representative approximations of mosquito densities, mapping areas with high increases in pupal productivity, and linking other environmental, entomological, or socio-demographic determinants, providing essential information to generate innovative strategies for prevention, vector control, and management of dengue.We suggest others make their data available as well.

CONTEXTA
total of 160 biological records (3,806 specimens) of the immature stage (pupae) of Ae. aegypti were collected from dwellings in the municipalities of Patía (El Bordo), Miranda, and Piamonte, within the department of Cauca, in the southwest of Colombia.These municipalities are part of an ongoing research project titled "Spatial stratification of dengue based on the identification of risk factors: a pilot trial in the department of Cauca" headed by the Entomology Group of the National Institute of Health, the Secretary of Health of Cauca, and the University of Applied and Environmental Sciences (UDCA).Among other objectives, this project seeks to evaluate the robustness of the link between entomological variables, which explains the spatial pattern observed in the distribution of dengue fever.The data were collected between 2021 and 2022 by a multidisciplinary team of environmental health technicians, geographic and environmental engineers, and professionals with extensive experience in medical entomology.In this dataset, we distinguish three sampling periods for each municipality.For Patía: March and October 2021, and March and April 2022; for Miranda: April and November 2021, and May 2022; and for Piamonte, July 2021, and February and July 2022.A fourth sampling period was performed in August and October 2022 in the municipalities of Patía and Miranda, respectively, to increase the sample size to carry out the serotyping of the specimens.

Figure 1 .
Figure 1.Interactive map of the georeferenced occurrences hosted by GBIF[18].https://www.gbif.org/dataset/a51b1fb9-6e67-42b9-879b-5f21e4d85642 2.6; 2/3 c.f.: 3.0; 3/3 c.f.: 3.5.Female pupae productivity (Number of pupae assuming a 1:1 sex ratio at emergence) Estimate the number of emerging adult mosquito females produced per container.Pupae-per-person index (Pupal productivity/total population of the screened houses) Generate an estimate of the number of pupae per person in the screened household.Breteau Index (Number of containers with any pupae *100/Number of screened houses) Defined as total number of positive containers per 100 households inspected.Determined by a ratio of positive containers to screened households.
2 ml vials with RNAlater © for subsequent processing with molecular biology techniques at the Entomology group and the Genomics of Emerging Microorganisms group of the National Institute of Health (Bogotá, Colombia) [21].
05) was performed to detect the distribution of variables related to spatial clustering.The Moran Index ranges from −1 to 1, where −1 indicates dispersed clustering patterns, 0 indicates randomness, and 1 suggests perfect association.Next, a Local Moran Index analysis distributed the significant (p = 0.05) clusters of dwelling blocks into four types of local spatial association.(I) high(x)-high(y) indicates areas with high values of the variable (x) surrounded by values above the mean of the variable under analysis (y); (II) low(x)-high(y) indicates areas with low values of the variable (x) surrounded by neighbouring areas with values above the mean of the variable (y); (III) low(x)-low(y) indicates areas with low values of the variable (x) surrounded by areas with values below the mean of the variable (y); and (IV) high(x)-low(y) indicates areas with high values of the variable (x) surrounded by areas below the mean of the variable (y) [13].

Figure 2 .
Figure 2. Bivariate local Moran index for the variables pupal productivity (x) and number of adult mosquitoes (y) in the municipalities of Patía, Miranda, and Piamonte (2021-2022).

Table 2 .
[24,25]tive entomological measures by sampling locality.The total of positive screened houses for each mosquito species is shown as a total and as a percentage for each municipality[24,25].

Table 3 .
Results of the regression analysis and the bivariate spatial autocorrelation for entomological variables of Ae. aegypti in Patía, Miranda, and Piamonte, according to the seasonality of sampling.PPP: pupae per person, TWS: tanks for water storage, * p ≤ 0.05.