Introduction, establishment, and distribution of Aedes aegypti and dengue in a temperate capital of Brazil: a retrospective surveillance-based study

Summary

Background

Dengue is spreading to southern latitudes in Brazil, where the temperate climate was once a barrier to the primary vector, Aedes aegypti. In this study, our objective was to reconstruct the introduction, establishment, and subsequent expansion of Ae. aegypti and dengue in Porto Alegre, the southernmost state capital of Brazil, located in Rio Grande do Sul state.

Methods

This ecological study used entomological and epidemiological surveillance data and official reports obtained from municipal health authorities of Porto Alegre, from 2001 to 2021. Descriptive analyses were employed, supplemented by space-time scan statistics to identify high-risk vector abundance and dengue case clusters.

Findings

Ae. aegypti was first detected in Porto Alegre in 2001, spreading citywide by 2016. The first autochthonous dengue case was recorded in 2010, and by 2021 the disease was detected in 78% of the neighbourhoods. DENV-1 was the dominant serotype and most cases occurred among people aged 20–59. Clusters of vectors and dengue cases were more frequent during summer and autumn, but a few were also identified during winter. High-risk clusters for vectors were more frequent in the Partenon and Northwest regions and for dengue in the East, Centre, Partenon and South.

Interpretation

Ae. aegypti successfully established and spread within a temperate city in Brazil. The presence of vectors, a susceptible population and socio-environmental characteristics conducive to mosquito proliferation resulted in autochthonous transmission of dengue fever after the continuous introduction of imported cases. The climatic barrier to dengue transmission in the south of Brazil has shifted southward, coinciding with the colonisation of Ae. aegypti and the emergence of dengue in recent years in Porto Alegre.

Funding

Generalitat de Catalunya, European Commission, and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).

Research in context.

Evidence before this study

The spread and incidence of dengue is rapidly increasing across the globe. In the southern region of Brazil, where a temperate climate predominates, many cities have recently experienced dengue emergence. However, knowledge about the disease dynamics in this region is still limited.

In July 2024, we conducted a search on PubMed for studies related to dengue and the Aedes aegypti vector in southern Brazil and other temperate areas of South America. Most studies analysed dengue and Ae. aegypti in Argentina. These studies showed that autochthonous dengue transmission started years after the first recorded vector, and highlighted the impact of the constant introduction of imported cases on maintaining local transmission. For Brazil, one study warned about an increase in dengue incidence and the suitability for Ae. aegypti in the country with a discussion about the southern region. Other studies in the Rio Grande do Sul and Paraná, states of the southern region, confirm the presence of Ae. aegypti and local dengue transmission in many cities. However, none of them describe the introduction, establishment and expansion of Ae. aegypti and dengue in the south. Therefore, a study aiming to reconstruct the history of dengue and its vector in a temperate climate region in Brazil, distinct from tropical areas where dengue is already widespread, could provide valuable insights into the southward advancement of this disease in the country.

Added value of this study

Our findings emphasise that Ae. aegypti mosquitoes can successfully colonise and disperse in temperate regions like urban areas in southern Brazil. In Porto Alegre, since the mosquito was first detected in 2001, it spread across the city. Despite this, Porto Alegre remained free of local dengue transmission for another nine years. The identified high-risk clusters of mosquito infestation and/or dengue showed heterogeneous distribution in space and time, suggesting that characteristics of urban landscapes shape the disease and vector distribution. Importantly, the identification of high-risk clusters even during winter, which was highly unexpected considering that the city's climate used to be unsuitable for dengue, indicates sustained local transmission.

Implications of all the available evidence

The high-risk area identification enables the municipality to develop more targeted control strategies and, thereby, optimise resource allocation. At the same time, our study provides a foundation for further research in the field, including the development of predictive models. It is important to note that dengue emergence is not immediate after the mosquito colonisation. This means the process may be ongoing in other locations where competent vectors are present, and there is a flow of humans from endemic to naïve areas that could carry the viruses. This highlights the risk of dengue emergence in cities with a climate and urbanisation level similar to Porto Alegre and the importance of continuous vector monitoring and effective control.

Introduction

Dengue is a mosquito-transmitted viral disease that has increased exponentially in recent years. According to the World Health Organization (WHO), dengue cases increased from 505 thousand in 2000 to 6.5 million in 2023 worldwide.¹ Since 2022, Brazil has experienced epidemics with record-breaking counts of cases, partially a consequence of the disease spreading to previously unaffected areas, notably in the South region.² In 2024, the country experienced its worst dengue epidemic ever registered, with over 6.5 million cases reported by Oct 18, 2024.³

In Brazil, the main dengue vector is Ae. aegypti, a climate-sensitive mosquito that is highly adapted to urban settings and bites during the day. The tiger mosquito Ae. albopictus is also present in the country,⁴ and a previous study confirmed the presence of the dengue virus (DENV) in this species in a rural area.⁵ However, there is limited knowledge about the importance of Ae. albopictus in dengue epidemiology in Brazil.

Changes in climatic conditions and rapid unplanned urbanisation can contribute to Ae. Aegypti colonisation in new locations.^6,7 The mosquito is widespread throughout the Brazilian territory, including the southern region. This region used to be associated with absent or low Ae. aegypti infestation because of its temperate climate. The recent mosquito colonisation in such locations indicates that temperate climate zones are proving to be suitable for Ae. aegypti, as observed in other countries in the Southern Cone.⁸

Dengue distribution is associated with Ae. aegypti vector presence. However, for local transmission to occur, other conditions are necessary, including favourable ecological and climatic factors, immune susceptibility to disease in the local human population, and virus circulation.^9,10 In the absence of these conditions, Ae. aegypti populations can persist without sustained arbovirus transmission for extended periods, a phenomenon called “Aegyptism without arboviruses”.¹¹

In favourable settings, imported cases can lead to autochthonous transmission and outbreaks, as detected in cities such as Córdoba and Santa Fé (Argentina).^8,12 In Brazil's southern region, many cities have experienced dengue emergence and outbreaks in recent years,² including Porto Alegre.

Porto Alegre, the capital of Rio Grande do Sul state, is an important Brazilian metropolis with high connectivity through air, road, and water transport. Due to its recent history of dengue emergence, this region presents a valuable study area for generating knowledge that may aid in understanding the dynamics of the disease. It also may exhibit an eco-epidemiological context that is significantly different from other tropical capitals where dengue has been established for many years. Using cases and mosquito surveillance data from the Municipal Health Secretariat of Porto Alegre, this study aims to reconstruct the introduction, establishment, and subsequent expansion of Ae. aegypti and dengue at intra-municipal level over 20 years.

Methods

Study area

Porto Alegre is the capital of Rio Grande do Sul state in the south of Brazil. The municipality is divided into 83 neighbourhoods and 17 regions (Refer to Figure S1 in the Supplementary Material for details). It has an area of 495.39 km², of which 214.91 km² is urban. The population in 2022 was 1,332,845, resulting in a population density of 2690.5 inhabitants/km². The municipality has an average altitude of 10 m above sea level and a temperate climate with well-distributed yearly rainfall. The average annual temperature is 19.9 °C, ranging from 14.9 °C during winter to 24.6 °C in summer. The annual rainfall is approximately 1495 mm, with the most rainfall falling in winter (414 mm).¹³

Data

We obtained epidemiological and entomological surveillance data from the Health Secretariat of Porto Alegre, the human population size, and epidemiological and management official documents published by Porto Alegre (Fig. 1a).

Fig. 1.

(a) Diagram of the studied dataset and main analysis; (b) Total reported and confirmed dengue cases in Porto Alegre (RS) by epidemiological week from 2010 to 2021; (c) Weekly average of Ae. aegypti females captured by MosquiTRAPs between September 2012 and December 2021.

Epidemiological and management official documents

We collected additional data from epidemiological and management official documents from the Health Secretariat of Porto Alegre. By searching for the terms “Aedes”, “arboviruses”, “dengue”, “larval survey”, and “LIRAa”, we found information on the dates and locations of the first recorded instances of the vector and dengue cases, as well as the infested neighbourhoods.

Entomological information before 2012 (place of the first Ae. aegypti records and the number of infested neighbourhoods) and data from LIRAa (Levantamento Rápido de Índices para Ae. aegypti) (2006–2016) were obtained in these documents. LIRAa is a monitoring method based on immature forms of Ae. aegypti (larvae and pupae) to obtain entomological indicators and investigate the distribution of the Ae. aegypti vector. One of the indicators obtained is the Building Infestation Index (Supplementary Figure S3), which provides the percentage of positive houses.¹⁴ An index less than 1 is considered satisfactory, between 1 and 3.9 indicates alert, and greater than 3.9 indicates risk.¹⁵ This methodology was used in Porto Alegre between 2006 and 2016 as part of the surveillance routine. Prior to this, simple larval surveys were conducted.

Epidemiological data

In Brazil, dengue is a mandatory disease for reporting to the Notifiable Diseases Information System (Sistema de Informação de Agravos de Notificação—SINAN) from the Ministry of Health. Surveillance is based on passive case detection, meaning only patients seeking health care with a suspected diagnosis of dengue can be detected and reported to SINAN. Dengue case data aggregated at the municipal level from 2001 to 2009 were obtained directly from the SINAN platform, where they are publicly available. For case definition details, please refer to the Supplementary Material (item 1.2).

From 2010 to July 2021, we obtained SINAN's individual-level dengue case data for Porto Alegre provided by the city's Health Secretariat. In Fig. 1b, the data are aggregated by epidemiological week. These data included but were not limited to, the cases' address information, age, whether the case was autochthonous or imported, and laboratory information. Residential addresses were georeferenced using the package ggmap (version 4.0.0) and Google's Geocoding API in R (version 2023.06.1) (R Core Team, 2022), and then grouped by neighbourhood.

Entomological data

As part of the entomological surveillance in Porto Alegre, the municipal Health Secretariat installed MosquiTRAPs to capture adult mosquitoes in strategic locations starting in September 2012. We obtained from the Health Secretariat the number of captured Ae. Aegypti adult females from September 2012 to December 2021 by week, which reflects the frequency at which the traps are monitored (Fig. 1c).

The MosquiTRAP is a device outfitted with a synthetic oviposition attractant (AtrAedes) designed to lure and capture gravid Ae. aegypti mosquitoes on a sticker card.¹⁶ The traps were positioned outdoors at fixed locations, sheltered from rain and sunlight, with a distance of approximately 250 m between them. The Health Secretariat of Porto Alegre chose the installation sites based on the occurrence of autochthonous dengue cases and high infestation of Ae. aegypti measured by LIRAa (Getúlio Dornelles Souza, General Coordination of Health Surveillance, CGVS, Porto Alegre, personal communication). The Health Secretariat of Porto Alegre also provided laboratory data for real-time PCR to detect dengue virus (DENV) in all captured Ae. aegypti adult samples (please refer to Supplementary Material, item 2.2 for more detailed information).

Human population

Population counts from census data (Brazilian Institute of Geography and Statistics, IBGE 2010) and projections (2011–2021) at the municipal level were obtained in the DataSUS platform, provided by the Brazilian Ministry of Health. For the neighbourhood level, we obtained data from the WorldPop unconstrained population estimates per 100-m pixel for Brazil from 2010 to 2021 (available at https://www.worldpop.org). Using QGIS (QGIS.org, 2023), we cropped the raster images to the shapefile of Porto Alegre and used zonal statistics to calculate the population of each neighbourhood by year. Shapefiles of the intra-municipal administrative divisions (neighbourhoods and regions) were sourced from https://prefeitura.poa.br/smpae/observapoa.

Analysis

We conducted descriptive analyses and employed SaTScan (also known as scan statistics) for spatiotemporal cluster identification of two different outcomes: Ae. aegypti infestation (2012–2021) and dengue cases (2010–2021).

Descriptive analysis

For the descriptive analyses, we calculated the entomological indicator Mean Female Aedes Index (MFAI) by dividing the weekly number of Ae. aegypti female mosquitoes captured in MosquiTRAPs by the total number of traps inspected in the same week.

We calculated the annual dengue incidence per 100,000 inhabitants for the municipality and stratified by age group. We also calculated the dengue incidence per 100,000 inhabitants by year and the incidence for the all study period for each neighbourhood.

Spatiotemporal scan statistics

For the spatiotemporal scan statistics, daily data were used considering a time aggregation of 7 days and each epidemiological year was analysed separately for each outcome (Ae. Aegypti infestation and dengue cases). In this method, a cylinder moves through the study area, varying the base (space) and height (time) to detect a spatiotemporal cluster based on its relative risk. The relative risk is defined as the ratio of the risk inside the cluster to the risk outside. The clusters were ordered according to the likelihood ratio, with the highest maximum likelihood being the most likely, i.e., with stronger evidence of being a high-risk cluster. Please refer to the Supplementary Material for detailed information and formulas (item 1.4).

In this study, a cluster is defined as a geographically limited region with a statistically significant higher risk of Ae. aegypti infestation/dengue cases than the remaining study area in a specific time window. To assess statistical significance, we performed 999 Monte Carlo simulations, and a cluster was deemed significant if the p-value was less than 0.05. Other limiting parameters were set to restrict the clusters' size and duration. For both outcomes, the minimum cluster duration was set to 21 days, based on the average extrinsic and intrinsic virus incubation period at 25 °C,¹⁷ and the maximum duration was set to 50% of the epidemiological year. Additional limiting parameters specific to each outcome are described below.

Spatiotemporal clusters of female Ae. aegypti infestation

The spatiotemporal permutation distribution model was applied to identify high-risk clusters of Ae. aegypti infestation using the geographic coordinates of each trap, the number of adult females caught by the trap, and the date of trap inspection. The limiting parameter of the cluster's maximum spatial size was set to 50% of the at-risk population. For this analysis, the numbers of all captured mosquitoes were considered as the at-risk population.

Spatiotemporal clusters of dengue cases

To identify high-risk clusters of dengue cases, we used a discrete Poisson model approach and the number of confirmed dengue cases among Porto Alegre residents aggregated by neighbourhood and date of symptom onset. The limiting parameter of the cluster's maximum spatial size was defined as 10% of the at-risk population (the number of inhabitants). Clusters were also limited to having at least two cases. None of the neighbourhoods were excluded from the analyses because all of them have a population proportion smaller than the maximum cluster's size.

We used SatScan software version 10.1 (http://www.satscan.org) for cluster identification. Maps were prepared using R (version 2023.06.1) and the packages sf (version 1.0-18), ggplot2 (version 3.5.1), and colorspace (version 2.1-1).

Ethical approval

The Federal University of Minas Gerais Research Ethics Committee approved this study, with CAAE 59597022.8.0000.5149.

Role of the funding source

The funders had no role in the study design, data collection, data analysis and interpretation, report writing, or publication submission.

Results

Introduction, establishment, and spread of Ae. Aegypti

According to the information extracted from official documents, the Ae. aegypti mosquito was first detected in Porto Alegre in 2001 in the Nonoai and Teresópolis neighbourhoods (South-Centre). By the end of the same year, the vector had already colonised 17% of the city's neighbourhoods. In 2009, the vector was detected in 50 neighbourhoods, and in 2016, in all of them.

Between 2006 and 2016, the Building Infestation Index was higher than zero, with the highest value recorded in 2013 (5.8) (Supplementary Figure S3).

In September 2012, the adult mosquito monitoring system was installed, consisting of 714 MosquiTRAPs distributed in 23 neighbourhoods of Porto Alegre. The monitored area and the number of traps installed expanded over time (Supplementary Table S1), reaching the minimum number of traps in 2013 and 2014 (n = 714) and the maximum in 2019 (n = 1438). Between September 2012 and December 2021, 136,877 specimens of Ae. aegypti females were captured in traps.

The average capture of Ae. aegypti females exhibited a seasonal pattern, decreasing in autumn/winter (March to September), reaching zero mosquitoes in some traps for a few weeks and none in all of them during epidemiological weeks 33, 35 and 36 of 2013 (August and September), in week 30 in 2016 (July), and in week 31 in 2021 (August), with an increasing in spring/summer (September to March) (Fig. 1c). From 2018 until 2020, there was a decrease in the MFAI of captured Ae. aegypti females compared to previous years (Fig. 1c) and a decline in trap inspections during certain months (Supplementary Figure S2).

Of the total number of captured Ae. aegypti females, 99.9% (n = 136,850) were grouped in samples and tested for serotype identification. All four dengue virus serotypes were detected. In 2014, two serotypes (DENV-1 and DENV-3) were co-circulating; in 2015, they were DENV-1 and DENV-4 (Supplementary Table S1). However, the positivity rate was low, ranging from 0 to 0.38%, and in most samples, a specific serotype was not detected.

In the spatiotemporal analysis, at least one cluster of high Ae. aegypti infestation was detected each year (Table 1 and Supplementary Table S2), with the area of high-risk clusters expanding over the years (Fig. 2). Most clusters started between January and May (summer and autumn). Starting in 2014, some clusters persisted until winter (Supplementary Figure S4). A few clusters also began during winter (June, July, or August) (Fig. 2). The most likely clusters in 2015 and 2018 were identified during winter in the Vila Ipiranga neighbourhood (Northwest region). The longest-duration clusters occurred in 2014, 2018, and 2021.

Table 1.

Summary of the characteristics of the clusters for Aedes aegypti, Porto Alegre, Brazil, 2010–2021.

Year	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021
N° cluster	2	8	15	14	9	15	9	13	10	10
Median (IQR)
Duration in days	21.0 (20.5–21.5)	37.5 (24–41)	27.0 (20–88.3)	62.0 (28–112.8)	27.0 (20–41)	41.0 (27–100.5)	41.0 (27–139)	27.0 (20–62)	20.0 (20–60.3)	34.0 (27–74.3)
N° of traps	29 (15–43)	50 (14–70)	8 (1.8–13.8)	25 (10.8–44.5)	58 (25–73)	6 (3.5–34)	72 (17–101)	24 (10–40)	44 (27.5–96.8)	28 (26.3–66)
N° observed Ae. aegypti female	74 (39.5–108.5)	175 (65–350)	60.5 (32.8–119.3)	64 (26–157.8)	92 (52–311)	105 (28–209)	199 (50–373)	46 (34–67)	95.5 (47.3–144)	123.5 (84.3–280.5)
Radius (km)	0.79 (0.4–1.2)	1.29 (0.5–1.6)	0.37 (0.1–0.7)	0.67 (0.4–1.2)	1.06 (0.8–1.5)	0.26 (0.2–1.1)	1.51 (0.8–2.2)	0.82 (0.5–0.9)	1.31 (0.9–1.9)	1.09 (0.9–1.3)

Fig. 2.

Aedes aegypti high infestation clusters in Porto Alegre, Brazil, 2012–2021. The colours indicate the clusters' start period and the number of the cluster is ordered according to the maximum log-likelihood ratio, with 1 being the most likely cluster. The grey-shaded areas represent the traps' location.

During the study period, areas in two city regions (Partenon and Northwest) showed a persistent high risk of Ae. aegypti infestation (Fig. 2), with cluster size varying over time (Table 1).

Introduction, establishment, and distribution of dengue cases

Since 2001, when Ae. aegypti was first detected in Porto Alegre, until 2009, all confirmed dengue cases in the city had been imported. During this period, the total number of cases was, on average, 28, but 2002 was an atypical year with 129 reported cases (Fig. 3a). The first autochthonous dengue cases were reported in May 2010 in the Jardim Carvalho neighbourhood (East region). From January 2010 to July 2021, 7124 suspected dengue cases were reported, of which 96% (n = 6651) were laboratory-tested, and 18.6% (n = 1325) were laboratory-confirmed for dengue. An additional 100 cases were confirmed based on clinical-epidemiological criteria, resulting in 1425 confirmed cases (Fig. 1c).

Fig. 3.

Imported and autochthonous dengue cases in Porto Alegre, Rio Grande do Sul, Brazil. (a) Distribution of autochthonous and imported dengue cases by year from 2010 to July 2021. (b) The year of the first record of autochthonous dengue; (c) Accumulated dengue autochthonous incidence by 100,000 inhabitants between 2010 and July 2021.

Since 2010, the average weekly proportion of confirmed cases was 14%. Notably, in epidemiological week 20 of 2019, the highest number of confirmed cases (n = 64) was recorded, constituting 47% of the reported cases (Fig. 1b). The majority of cases (n = 1395, 97.85%) did not present warning signs, and deaths were not recorded.

Of the confirmed cases (n = 1425), 1330 (93%) were Porto Alegre residents, 988 (74%) were classified as autochthonous (Figs. 3a) and 342 (26%) were imported. The probable infection places for imported cases were distributed across 25 Brazilian states. The only two states for which no imported cases were recorded were Amapá and Acre. For 11% (n = 38) of cases, there was no information regarding the place of infection. Most infections occurred in the states of Rio de Janeiro (n = 90) and São Paulo (n = 39). Regarding the spatial distribution of imported cases in Porto Alegre, there is a concentration in the Central region of the municipality (Supplementary Figure S5).

Considering only the residential neighbourhoods of the autochthonous cases, the disease showed expansion, reaching 69% (n = 57) of the neighbourhoods by 2016 and 78% (n = 65) by the end of the study period (Fig. 3b). Over the years, dengue cases tended to accumulate in a few neighbourhoods of the municipality, mainly in those that were the first to report autochthonous cases (Fig. 3c).

During the study period, the lowest annual dengue incidence in the municipality was observed in 2018 and the highest in 2019 (0.07 and 31.87 cases/100,000 inhabitants, respectively; Supplementary Table S3). The regions with higher incidence were North, Northwest and East. For all years, the incidence rate among the neighbourhoods varied considerably (Supplementary Figure S6), notably in 2016, ranging from a maximum of 753 cases per 100,000 inhabitants in the Chácara das Pedras (East region) and a minimum of 3.5 cases per 100,000 inhabitants in Santa Tereza (Cruzeiro region).

High-risk clusters for dengue cases were detected in all years except 2017 and 2018 (Fig. 4, Supplementary Table S5). The minimum number of neighbourhoods per cluster was one, and the maximum was 11 (Table 2). The median duration varied from 23.5–111 days (Table 2). The longest cluster occurred in 2013 (ID2) and 2019 (ID1), both lasting 118 days. The majority of clusters started between January and April (summer and autumn) (Fig. 4 and Supplementary Figure S6). In 2010, one cluster (ID2) began during winter, and in 2019, one cluster (ID3) started in June and remained until July, while another (ID1) started in March and continued until the beginning of July (Supplementary Figure S7). We observed that the first dengue clusters often occurred in the East, Partenon and Centre regions.

Fig. 4.

High-risk clusters for dengue cases in Porto Alegre, Brazil, 2010–2021. The colours indicate the clusters' start period, and the cluster labels are ordered according to the maximum log-likelihood ratio, with 1 being the most likely cluster.

Table 2.

Summary of the characteristics of the clusters for dengue, Porto Alegre, Brazil, 2010–2021.

Year	2010	2011	2012	2013	2014	2015	2016	2019	2020	2021
N° cluster	2	2	2	3	1	3	6	3	2	3
N° of neighbourhoods	5	14	12	13	1	14	24	6	11	4
Population inside cluster	130,725	177,270	146,451	266,771	13,484	262,156	439,939	167,351	176,260	43,604
N° observed dengue cases	10	24	10	136	4	30	233	393	11	43
Median (IQR)
Duration in days	23.5 (21.8–25.3)	55.0 (51.5–58.5)	58.5 (46.3–70.8)	111.0 (93.5–114.5)	34.0 (34–34)	62.0 (55–62)	79.5 (60.3–83)	55.0 (44.5–86.5)	55.0 (37.5–72.5)	48.0 (34–55)
Relative risk	71.63 (53.3–90)	53.93 (39.2–68.7)	32.74 (30.4–35.1)	18.49 (12.5–19)	320.12 (320.1–320.1)	13.66 (11.7–43.8)	21.48 (5.3–39.1)	34.35 (27.8–77)	34.32 (23.8–44.8)	195.4 (166.3–277.9)
Radius (km)	1.09 (0.6–1.6)	1.89 (1.7–2.1)	2.02 (1.9–2.2)	2 (1.6–2)	0 (0–0)	2 (1–2.1)	2.43 (0.2–4)	0.85 (0.4–1.3)	2.44 (2.1–2.7)	0 (0–1)

The most likely clusters of all years included 26 neighbourhoods of the municipality, the most frequent being Chácara das Pedras (East region), Ipanema (South region), and Santo Antônio (Partenon region). During the study period, clusters occurred in 67 neighbourhoods. In general, clusters were more frequent in the East, Centre, Partenon, and South regions. In 2016, clusters were observed in a larger geographic area (24 neighbourhoods) compared to other years.

Age profile and viral serotype of confirmed dengue cases

Most confirmed cases were 20–39 years old, accounting for 38% of cases (n = 546) and in 40–59, with 32% (n = 451) (Supplementary Figure S8). The predominant age range did not vary over the years; however, in the years when transmission was more intense (2013, 2016, and 2019), all age groups were affected (Supplementary Figure S8).

The dengue annual incidence among residents, calculated for each age group, was highest among the 10–19 years old (42.6 cases/100,000) in 2019. In 2013 and 2016, the highest annual incidence was observed in the 20 to 39 age group (18.9 and 30.5 cases/100,000, respectively). The lowest incidence was recorded among children aged 0 to 4 and 5 to 9, except in 2021, when those aged 60 years or more showed the lowest incidence (Supplementary Figure S9).

During the study period, laboratory results on the DENV serotype were available for 17.8% of the laboratory-confirmed cases (253/1425). Of these, the serotype was identified in 94% of samples. DENV1 was detected in 65% of them, followed by DENV2 at 22% and DENV4 at 14%. DENV1 was detected in all years except for 2020, while DENV-3 was not detected. In 2013, DENV1, DENV2, and DENV4 serotypes co-circulated (Supplementary Table S4).

Discussion

Our study integrates entomological and epidemiological data over 20 years to describe at the local level the introduction and establishment of Ae. aegypti and dengue emergence in a temperate capital city in Brazil. While imported cases were consistently observed throughout the study period, primarily originating from the Southeast region of Brazil, we observed a significant shift after 2013, with a notable increase in autochthonous cases, marking an important transition in the disease dynamics. Importantly, we identified high-risk clusters even during winter, which were unexpected considering that the city's climate used to be unsuitable for dengue and indicates sustained local transmission.

Based on larval surveys, Ae. aegypti was first detected in Porto Alegre in 2001 and spread throughout the city by 2016. However, larval surveys have a low sensitivity for detecting the presence of the vectors.¹⁸ Therefore, it is possible that the vector was present sooner and dispersed more quickly than the surveillance could detect.

Since 2012, MosquiTRAPs have been used to monitor adult mosquitoes and generate the MFAI index, which showed fluctuations with peaks during the summer months and low values in winter. This pattern, which is associated with fluctuations in the mosquito population, is similar to what is observed in Córdoba,¹⁹ a temperate city in Argentina, and in other tropical cities in Brazil.^20,21

During the summer, Porto Alegre experiences more favourable temperature conditions for Ae. aegypti mosquitoes.²² Mosquitoes are driven by temperature because of their poikilothermic metabolism. An increase in temperature, up to a certain threshold, enhances mosquito activity, reduces developmental time, and raises adult survival rates,²³ leading to an increase in the vector population. Furthermore, warmer temperatures influence vector competence by accelerating viral replication and shortening the viral incubation period in mosquitoes,²³ elevating the dengue transmission risk.

Unlike many other Brazilian cities, where the vector population can be associated with increased precipitation during the summer, Porto Alegre registers rainfall throughout the year, including in the winter, when temperatures are lower.¹³ This represents a complex interaction of climatic factors and mosquitoes in the city.

The lower winter temperatures, average minimum temperature11 °C,¹³ are unfavourable for Ae. aegypti²⁴ and clearly impose restrictions on the mosquito. In some winter weeks, no mosquitoes were caught, which differs from other tropical areas that use the same type of traps.²¹ However, we observed a quick re-establishment of the MFAI after winter. This suggests that mosquitoes may be resisting the cooler winter temperatures in their egg phase.

Between 2018 and 2020, we noticed a decrease in the MFAI, which was partially explained by the deactivation of some traps due to logistical problems. Because of this, some neighbourhoods were not inspected, leading to an underestimation of the infestation. This emphasises the importance of continuous entomological surveillance, which is essential for designing effective vector control strategies.

Despite the mosquito's widespread presence in Porto Alegre, the abundance distribution in the monitored areas was heterogeneous (Fig. 2). The Partenon region was the most critical because of its persistence as a high-risk area for Ae. aegypti infestation. Most neighbourhoods in the Partenon region are characterised by poor households, high garbage accumulation in their surroundings, and a large part of the population living in slums.²⁵ There is evidence suggesting that inadequate sanitation, high population density, and low income are linked to a higher vector prevalence, creating conditions conducive to mosquito proliferation.^26,27 The persistence of a high infestation cluster for extended periods, as observed in this study, indicates challenges in controlling mosquito infestation.

The detection of high mosquito infestation and dengue clusters during winter suggests the persistence of the vector's presence in the city and sustained dengue transmission for extended periods beyond the typical season. Sporadic dengue cases during winter were reported in 2009, 2011, 2018, 2019, and 2020 in Santa Fé (Argentina), another temperate city.¹² This underlines the importance of maintaining vector surveillance and control even during the winter months.

Interestingly, Vila Ipiranga (Northwest region) has consistently been identified as part of a high infestation cluster since it began being monitored in 2015, and it was identified as part of the most likely clusters in 2015 and 2018 during winter. This neighbourhood has a high income per capita, waste disposal, water supply, adequate sanitary sewers, high literacy rates, and absence of people living in slums (Supplementary Figure S10). Its identification as a high infestation area is unexpected, and the possible factors favouring the vector in this region warrant investigation. With a large flow of people from other Porto Alegre neighbourhoods and surrounding municipalities, Vila Ipiranga may also be contributing to sustained dengue transmission for extended periods.

As the results indicate, the climate of the southern region did not prevent colonisation and (re-)establishment of the Ae. aegypti vector. Similar findings were observed in other temperate cities such as Córdoba and Santa Fé (Argentina) and Salto (Uruguay).^8,28 There are several factors that could explain the establishment of the vector in this region, such as gradual changes in climatic conditions, the adaptation of Ae. aegypti populations to lower temperatures, and the presence of breeding sites/shelters capable of maintaining temperatures higher than the environment (e.g. concrete tanks).²⁹ Notably, citizens from Porto Alegre frequently reported the existence of water tanks in private homes with conditions conducive to breeding sites for mosquitoes.³⁰

From the detection of Ae. aegypti in Porto Alegre to the first record of local dengue transmission, nine years had passed. Dengue emergence in temperate regions has become a more common phenomenon in the last decade.^2,31 In the Southern Cone of South America, only continental Chile has not recorded autochthonous dengue cases until mid-2024.³² It is interesting to note that, as observed in this study, the disease emergence in other temperate regions also occurred years after the vector detection (for the first time or due to reintroduction). In Córdoba and Salta Province, Argentina, dengue emergence occurred after 14 and 11 years of detection of the vector's presence,^33,34 respectively. In Uruguay, it happened after 19 years.^35,36 This raises concerns for other cities or countries where the vector is present and established, but local dengue transmission has not yet occurred. This characterises the “Aegyptism without arboviruses”, which is the case in Chile and many southern cities in Brazil.

Before the confirmation of autochthonous dengue transmission in Porto Alegre, imported cases were introduced multiple times in the city. The constant introduction of imported cases may have led to its establishment and spread, leading to persistent transmission.¹⁰ Since the emergence of dengue fever, the number of cases has gradually increased and in 2024, up to epidemiological week 31, there were 36,556 cases notified in the city,³⁷ more than five times the number of all suspected cases in the entire study period. Córdoba, Argentina, registered dengue emergence one year before Porto Alegre, and in the same period, reported 127,490 cases.³⁸ These figures highlight that after the establishment of the disease, controlling its spread is a significant challenge.

Dengue outbreaks in Porto Alegre occurred at the same time as major epidemics in the Southeast region,³⁹ where most imported cases were infected. Therefore, if the same pattern of flow of imported cases is sustained, it is advisable to maintain syndromic surveillance of people coming from important tourist cities in this region, such as Rio de Janeiro.

Within Porto Alegre's territory, a heterogeneous distribution of dengue cases was observed. Some neighbourhoods appeared frequently in areas of high-risk clusters, and their socioeconomic and demographic indicators vary, with distinct social vulnerability (Supplementary Figure S9). For example, Chácara das Pedras has high socioeconomic levels and good environmental conditions, while Ipanema and Santo Antônio have lower per capita incomes and greater garbage accumulation in their surroundings. Studies demonstrate that the association between poverty and dengue depends on the locality and period of study.^40,41 This association likely changes depending on the epidemiological moment, since population immunity may largely impact the disease distribution and is not taken into account in most studies.^42,43

Interestingly, the years with the highest number of dengue cases were preceded by the longest high-vector infestation clusters. In 2018, a high vector infestation cluster of 181 days (Supplementary Table S2) occurred in the same region where a dengue cluster was detected in 2019 (Figs. 2 and 4). This indicates that the identification of high infestation areas may help guide interventions to prevent or reduce dengue cases in the following year.

Frequently, the most likely cluster of the vector and dengue in each year occurred in contiguous areas. Spatial overlap of the most likely clusters of vectors and cases happened in the Partenon neighbourhood, indicating that the area is critical for vector infestation and disease transmission. However, we identified dengue clusters in areas without installed traps and high-risk vector infestation clusters in areas with no records of dengue cases (Supplementary Figure S11). This raises a discussion about the optimal distribution of traps and the possibility of disease spread to unaffected neighbourhoods.

Considering integrated surveillance, mosquito and human assays to detect DENV are part of the routine in Porto Alegre. However, our findings indicate a low virus detection rate and identification of serotypes in the mosquito population. We also observed a low correspondence between circulating serotypes in humans and mosquitoes, which could be partly due to the limited number of human samples tested. This leads to a loss of valuable information for tracking the introduction or changes in the circulating serotypes, with significant epidemiological implications.⁴⁴ It is likely that improvements in the methods used to process mosquito samples collected in the field may enhance the sensitivity of laboratory serotype testing. Additionally, we recommend increasing the percentage of human case samples tested for serotype identification.

The predominance of cases aged 20–39 and 40–59 was consistent with previous studies in Brazil.³⁹ This age group comprises the economically active population and can cause a significant impact on the economy.⁴⁵ Changes in the age profile of dengue cases can indicate the introduction of a new serotype⁴⁶ or a change in the disease endemicity.⁴⁷ However, we did not observe this trend in Porto Alegre. It is important to highlight that we have presented results from a series of 12-year local transmission and tracking and monitoring this age profile is important, especially in a city like Porto Alegre, where the data has been well documented since dengue emerged, and because it is located in the southern region of Brazil, where the increase in dengue transmission is relatively recent.

It is important to consider the limitations of this study. First, we did not have adult mosquito data for the entire study period, as before 2012 the mosquito surveillance was conducted solely through larval surveys. Adult Aedes data was not available for the entire city, as it was conditioned on the position of the traps installed by the Health Secretariat of Porto Alegre. Additionally, variations in the number of weekly inspections carried out by endemic control agents may have led to suboptimal sampling. Furthermore, the type of mosquito traps used in Porto Alegre has demonstrated lower performance in detecting seasonal vector population variation compared to other traps used in other localities.^18,48 As national dengue surveillance is passive, asymptomatic cases cannot be captured. Finally, the COVID-19 pandemic impacted both epidemiological surveillance and entomological monitoring, especially in 2020, with fewer people seeking health care, the inactivation of several traps, and a decrease in the frequency of trap inspections.

Despite the limitations, this study provides a robust analysis at a high spatial and temporal resolution, integrating entomological and epidemiological surveillance data, and complementing the information with available management and epidemiological reports. Our study's strengths also rely on the quality and availability of the entomological data, which is often scarce in Brazil, particularly when considering the historical series of weekly monitoring of adult mosquitoes spanning nearly a decade. Continuous vector monitoring data and dengue case reports since the emergence of local transmission provide valuable insights into the geographical and temporal trends of Ae. aegypti infestation and dengue cases. The high-risk area identification enables the municipality to develop more targeted control strategies and, thereby, optimise resource allocation. We acknowledge, however, that the descriptive nature of this study leaves important questions unanswered, particularly those related to the mechanisms underlying the observed phenomena. Notably, the associations between socioeconomic, demographic, and climatic variables and the spatiotemporal distribution of the disease and its vector were not statistically assessed. Nonetheless, we believe that our findings offer a valuable foundation for future research in the field, including the development of hypothesis-driven analyses and predictive models.

Concurrently, the vector introduction, establishment, and expansion process, with local dengue transmission emerging many years later, serves as a warning for other locations where the vector is present but local transmission has not been confirmed yet. This underscores the importance of continuous vector monitoring and effective control, even in the absence of local transmission.

Contributors

DACF participated in the study's conceptualisation, data curation, formal analysis, and wrote the original manuscript draft. LPF participated in the interpretation and presentation of results, and write-review & editing. RL participated in the presentation of results, and write-review & editing. GDS participated in the data curation, interpretation of results, and write-review & editing process. RTF participated in the write-review & editing process. RML participated in the conceptualisation of the study, data curation, interpretation of results, write-review & editing and supervision. DACF and RML had access to raw data and were responsible for the decision to submit for publication.

Data sharing statement

The entomological data are owned by the Municipal Government of Porto Alegre. Requests for this data should be made directly to this entity. The epidemiological data were provided with the authorisation of the ethics committee, and we do not have permission to share them. The remaining data used are in the public domain and can be accessed directly through the sources indicated throughout the article.

Editor's note

The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors used ChatGPT and Grammarly to check the grammar and edit the language of the original text created by the authors.

Declaration of interests

The authors declare no conflict of interest.