Introduction

Despite decades of coordinated mosquito control efforts ranging from mechanical, chemical to integrated mosquito control in the U.S. and internationally, mosquitoes remain the most common disease vectors globally1,2,3,4. Mosquitoes are responsible for spreading a wide range of pathogens of public health importance such as plasmodium falciparum, dengue, Zika and West Nile viruses, that can have severe and life-threatening consequences (e.g., encephalitis, meningitis, hemorrhagic fever, and microcephaly)5,6,7. Different risk factors influence the resurgence and spread of mosquito-borne diseases, mosquito abundance, species richness, and abundance of susceptible vertebrates: meteorological factors (e.g., temperature, relative humidity, and precipitation), urbanization, sociodemographic factors, ecology, and changes in public health policy8,9,10,11,12,13,14,15. In coastal cities, recent studies have indeed linked the spread, activity, and longevity of vector mosquitoes to climatic changes16. These changes can have significant implications for mosquito population dynamics, disease transmission, and public health preparedness. Climate change, including rising temperatures, altered precipitation patterns, and sea level rise, affects mosquito habitats, breeding sites, and behavior11,17,18,19 As a result, coastal cities face increased risks related to mosquito-borne diseases, such as dengue, Zika, and West Nile (WNV)20,21,22,23,24. Understanding these connections is crucial for effective mosquito control strategies and disaster response planning25,26,27,28,29,30,31,32.

Hurricanes bring significant abiotic changes to mosquito populations11,26,33. Favorable effects include increased breeding sites due to heavy rainfall, nutrient enrichment in floodwaters, warmer temperatures, and elevated humidity34,35. However, unfavorable changes include habitat disruption and salinity fluctuations36. Understanding this complex interplay is crucial for effective mosquito control and public health preparedness.

Previous studies in the U.S. have documented important relationships between mosquito activity and post hurricane events (increases in the mosquito Culex quinquefasciatus in New Orleans, Louisiana following Hurricane Katrina)11 and increases in Aedes aegypti populations (e.g., five weeks after Hurricane Maria in Puerto Rico)37. Similar results were reported in Collier County, Florida, where daily landing rates of 150 adult females/two-minute period were observed for the black salt-marsh mosquito Aedes taeniorhynchus following Hurricane Irma but decreased a year later38. In Texas, Cx. nigripalpus increased following Hurricane Harvey39. Weather-related factors such as temperature, relative humidity and precipitation were reported to influence mosquito seasonal activity, as they can alter mosquito development and reproductive and mortality rates40,41,42. However, elsewhere (Louisiana), the results of mosquito activity in the context of hurricanes and storm events have been mixed43, with some studies conducted after Hurricane Katrina in Louisiana and Mississippi finding increases in the number of reported cases of mosquito-borne disease (West Nile neuroinvasive disease) compared to previous years44 and an increase in the average total number of infected mosquitoes (specifically WNV) on the northern coast of the Gulf of Mexico36. The findings of other studies contradict these findings43 or revealed no arboviruses in mosquitoes collected six weeks after Hurricane Katrina34.

A recent global scoping review on flooding events and mosquito-borne diseases noted that although flooding is linked to an increased incidence of mosquito-borne diseases (dengue and malaria), the results were mixed regarding lag periods, with studies focused on dengue reporting short-term (less than a month) decreases and subsequent (1–4 month) increases in incidence, while those focused on malaria noted post event incidence increases; however, the results were mixed, and the temporal pattern was less clear45. Notably, such methodological inconsistencies limit direct comparison and generalizability of the study results.

Although the mosquito-borne diseases compared here are vectored by different mosquitoes with unique ecology and caused by entirely different pathogens (virus vs Protozoa), in the hurricane-prone state of Florida, mosquito-borne arboviruses (WNV, Saint Louis encephalitis virus, and Eastern equine encephalitis virus), and increases in the mosquito population have been reported post storm, posing a major threat to public health38,46,47,48. Notably, of the 292 hurricanes of which landfall occurred in the continental U.S. between 1851 and 2018, 123 (41%) struck Florida49, and projections show an increasing number of major hurricanes globally50, suggesting that coastal cities such as Miami, Florida, are at heightened risk from vector and nuisance mosquitoes. Moreover, rising sea levels may combine with local climate factors (e.g., temperature, rainfall, and wind) to increase the transmission of mosquito pathogens-causing diseases17 by influencing their vectors51,52.

Notably, the potential effect of rising sea levels on the prevalence of vector-borne infectious diseases has been documented by Ramasamy and Surendran17,52. Briefly, global climate change, exacerbated by rising sea levels, poses a significant risk to the transmission of mosquito-borne diseases in coastal zones. For example, as saline and brackish water bodies expand due to warming oceans, salinity-tolerant mosquitoes thrive, while freshwater mosquito vectors adapt to the changing environment. Therefore, monitoring disease incidence and developing preventive strategies are crucial to mitigate these risks. Furthermore, extreme climate events such as floods may increase the risk of vector-borne disease spillover53. Despite this, the potential impact of hurricanes on mosquito species composition or the window of lag time required for potential vector populations to develop and recover post-storm in Miami-Dade County is yet to be determined.

To address this gap in knowledge, first, we assessed adult mosquito populations (all species) before and after Hurricane Irma to understand the impact of storm events on mosquito abundance and estimated the time lags required for population development and recovery. Second, we determined which mosquito species predominated in the aftermath of Hurricane Irma. This information is crucial for understanding vector populations and their dynamics following extreme weather events. Third, we evaluated mosquito species composition using different trap types (e.g., light traps, gravid traps, CO₂ traps). By comparing trap-specific captures, we provide insights into species preferences and behavior during storm-induced disturbances in Miami-Dade County, Florida.

Methods

Study area

Miami-Dade County is in the southeastern part of the State of Florida and is the southeasternmost county on the U.S. mainland, between longitude − 80° 11′ 22.20" W and latitude 25° 46′ 16.19" N. Approximately 56% of Miami-Dade County’s land is 6 feet or less above sea level (this county is commonly called Miami, Miami-Dade, Dade County, Dade, Metro-Dade, or Greater Miami). Miami-Dade County is the third largest county in Florida, it has a surface area of 5,040 km49 (1,946 mi49). The average annual rainfall in the county is 61 inches of rain per year (7 inches more than the annual average in Florida), and the maximum and minimum average monthly temperatures are 90.6 °F and 41.8 °F, respectively. Peak rainfall generally occurs from mid-May through early October. In 2022, 12% of Florida’s total population were home to Miami-Dade County (2,701,762 population), which is one of the largest and most ethnically diverse yet poorest counties in the US11. The county was the epicenter of the 2016 Zika outbreak13, and cases of locally transmitted human cases of West Nile virus and dengue virus were diagnosed as recently as 202354,55.

Data sources

This was a retrospective observational study involving mosquito surveillance data and meteorological, socioeconomic, and demographic data. Table 1 presents the data variables of interest, spatial and temporal scales, and corresponding processing.

Table 1 Summary of the input data considered for inclusion, Miami-Dade County, Florida.
Full size table

Mosquito data collection

Adult mosquito surveillance data from 2016 to 2018 were provided by the Miami-Dade County Mosquito Control and Habitat Management Division or MDC-MCHMD (from August 15, 2016, to October 15, 2018). The MDC-MCHMD uses two trap types; CDC light traps (John W. Hock Company, Gainesville, FL, USA) located at 30 sites and baited with dry ice and Biogents Sentinel (BG-Sentinel) traps located at 131 trap sites (Fig. 1). CDC light traps are sparsely and permanently located in remote and wooded areas across the county, while BG-Sentinel traps are emptied and rotated among the sampling sites. While trapping in CDC light traps has a long history in the county (established in 2013), the analysis was constrained to mosquito capture data from 2016 to 2018 to align with mosquito data collected in BG Sentinel traps (established in 2016 after the Zika outbreak) in the study area. The collected mosquitoes were processed and morphologically identified to species at the MDC-MCHMD laboratory using taxonomic keys56.

Fig. 1
figure 1

Source: Miami-Dade Mosquito Control and Habitat Management Division, MDC, Florida, 2018. Map generated by the first author using ArcGIS software version 10.31

The spatial distribution of mosquito trap locations by trap type: (a) gravid traps and (b) light traps in Miami-Dade County, Florida. 61

Full size image

Study period and epidemiological weeks

We used weekly mosquito data collected from active mosquito trap types (CDC light traps and BG-Sentinel traps) collected within 24 h from Monday to Tuesday every week between 2016 and 2018. Because one objective of this study was to compare mosquito species composition and relative abundance pre- and post-Hurricane Irma in Miami-Dade County, we aggregated mosquito counts according to CDC week or epidemiological week (epi-week), a standardized method of counting weeks, to allow for comparisons of data year after year57. Since epi-weeks change annually and because Hurricane Irma struck Miami-Dade County on September 10, 2017, as a Category 3 storm, the mosquito data used in the analysis were further constrained to 8 epi-weeks.

The first four epi-weeks comprised the pre-hurricane Irma period corresponding to the dates August 15 to September 6, 2016, and the post-hurricane Irma period, which comprised the dates September 19 to October 15, 2018 (Table 2). Epi-week 37 or the week corresponding to Sep 12–18 in each study year was excluded from the analysis because this week denotes Hurricane Irma’s landfall in Miami-Dade County, and traps were temporarily taken down during that week. Mosquito data for 2016 and 2018 were used as baselines.

Table 2 Number of active mosquito traps during the study period (epi-week 33–41; August 15, 2016-Otober 2018), Miami-Dade County, Florida.
Full size table

The number of active traps used was divided by the number of mosquitoes (count) by the traps count to yield the mosquito/trap/night (Table 3). This data collection process yielded 609 trap nights pre-Hurricane Irma (n = 489 trap nights for BG-Sentinel traps and n = 120 trap nights for CDC light traps) and post-Hurricane Irma; 639 trap nights were estimated (n = 524 trap nights for BG-Sentinel traps and n = 115 trap nights for CDC light traps).

Table 3 Total mosquito captures per trap type and trap night, Miami-Dade County, Florida.
Full size table

Socioeconomic and human demographic variables

Because sociodemographic factors have been linked to mosquito-borne disease risk, sociodemographic data were downloaded from the U.S. Census Bureau American Community Survey (ACS) 5-year estimates (2013 to 2018). This period was used as a better representation of the baseline (2016, 2018) and for comparison with the mosquito data (2017). The variables of interest are similar to those used in previous similar studies in Chicago, Detroit, Georgia, and New Orleans11,58,59,60 and include population density, percentage population Hispanic, African American, gender, median household income, education attainment (a bachelor’s degree or higher), and unemployment.

Meteorological variables

We obtained historical daily mean temperature and rainfall data reported at the Miami International Airport (MIA) Station for an equivalent study period (August 15, 2016, to October 15, 2018). The data included weekly maximum temperature, weekly minimum temperature, and weekly rainfall. We also obtained rainfall raster files for September 10 to 12, 2017, to represent the Hurricane Irma period from the National Weather Services Advanced Hydrologic Prediction Service in GeoTIFF format.

Administrative boundaries

Administrative and evacuation zone boundaries at the census tract level for 2010 were obtained from the Miami-Dade County GIS Hub (https://gis-mdc.opendata.arcgis.com/). The evacuation zone is divided into five risk levels, A to E, with A being the most vulnerable.

Data analysis

Data processing

Only female mosquito counts from the two study trap types were used in the analysis since only female mosquitos need blood meal for egg development. Given that the mosquito capture data was highly skewed, we used log transformation to reduce the skewness and stabilize the variance. Summary statistics were compiled by trap night for each study epi-week to denote the pre-post Hurricane Irma periods, with results presented for both epi-week variation and variation independent of year. We then geocoded all mosquito data from the identified active traps captured in both CDC and BG-Sentinel traps (n = 161 traps; 30 CDC light traps and 131 BG Sentinel traps) using ArcGIS version 10.361. Of the 161 mosquito traps maintained by the Miami-Dade County Mosquito Control and Habitat Management Division, 96 BG-Sentinel traps and 28 CDC light traps were active during the study period. The geocoded mosquito data were then assigned to their appropriate census tracts along with demographic and socioeconomic variables. To facilitate comparison over time, an average was used when multiple traps existed in a census tract. Of the 519 census tracts in Miami-Dade County, 13.7 (n = 71) were active before and after Hurricane Irma.

To extract the rainfall raster data points corresponding to the mosquito traps, we applied a focal statistics tool in ArcGIS version 10.3161 to derive mean rainfall values for each census tract. In ArcGIS, the ‘focal statistics’ tool calculates statistics for a focal cell within a defined moving window. For each input cell location, this tool computes a statistic based on the values within a specified neighborhood around it. To achieve this, it uses the maximum, minimum, and mean values of a cell from its neighboring cells. Users can directly calculate these values by specifying the required parameters in the ‘focal statistics’ window. In the context of this study, we calculated an input cell location and recorded a statistic of values within a specific census tract. This information was then spatially joined to the study points (mosquito traps) and appropriate census tracts for statistical analysis.

Statistical analysis

Having confirmed that the assumptions for analysis of variance (ANOVA)—normality, homogeneity of variances, and independence—were met during exploratory data analysis, we proceeded to perform an ANOVA. Our goal was to evaluate whether mosquito abundance in CDC light traps and BG-Sentinel traps differed before and after Hurricane Irma. The dependent variable was the number of adult female mosquitoes per trap at night by trap type. Independent variables of interest were the different mosquito species (categorical variable) and a binary dummy variable that depicted Hurricane Irma (coded as 0 = pre- and 1 = post-Hurricane Irma). Included in the analysis were 20 species captured in light traps and 13 species captured in BG Sentinel traps. These species are of public health importance by the Miami-Dade County Mosquito Control and Habitat Management Division. Analysis was performed using IBM SPSS Statistics for Windows Version 25.062.

Ordinary least squares (OLS) and geographically weighted regression (GWR)

We applied both ordinary least squares (OLS) and geographically weighted regression (GWR)63 models to assess the effects of sociodemographic and meteorological factors (temperature, rainfall, and sociodemographic data) pre- and post-Hurricane Irma and across census tracts in Miami-Dade County on mosquito abundance (average number of mosquitoes in each census tract per week). The detailed information and spatial distributions of the dependent and independent variables used for OLS and GWR are shown in Table 1. OLS was applied to evaluate the global relationships between the dependent and independent variables. Given the spatial heterogeneity of our study area Miami-Dade County and because OLS regression assumes that the data are normally distributed and that the regression coefficients are “global” and apply equally across the entire study area and that residuals are spatially random, this can bias the OLS scores and inflate their significance64. Specifically, we observed that residuals for the OLS regression model (via Moran’s I, a diagnostic statistic)65,66,67 were randomly distributed for mosquitoes captured pre–Hurricane Irma but were dispersed for those captured post-Hurricane Irma, which violates the assumption of OLS regression68.

Given these findings, we used GWR to help identify the spatial influence of census tracts/traps, which are not explained by OLS regression. We used cross-validation (CV) as the bandwidth method and opted for the adaptive kernel type in the analysis settings for GWR because the distribution of traps/census tracts was heterogeneous in the Miami-County study area. To compare the model performances of OLS and GWR, we used the coefficient of determination (R2) and corrected Akaike’s information criterion (AICc)63,69. To evaluate the extent of multicollinearity among the independent variables (meteorological correlates, temperature, rainfall, and sociodemographic factors) before and after Hurricane Irma, we used the variance inflation factor (VIF) for the set of all nine covariates prior to model fitting, where VIF values greater than 10 indicates problematic collinearity/ redundancy among the variables70. Multicollinearity was not indicated (values were less than 10) in full models, and all variables were used in the model71. The GWR and OLS analyses were conducted using GWR software (version 3.0; Newcastle University, England, UK).

Results

Mosquito abundance and composition by week and year

Between 2016 and 2018, a total of 535,095 adult female mosquitoes comprising 32 mosquito species were captured on 1,248 trap nights in the two trap types. Of the trap nights, 609 trap nights were conducted before Hurricane Irma, and 639 trap nights were conducted after Hurricane Irma (Table 4). The mosquito abundance was 5.3 times greater in 2017 (406,842) than in 2016 (77,237) and 8.0 times greater in 2017 (406,842) than in 2018 (51,017). Cx. nigripalpus represented 70.4% of the 535,096 mosquito captures, followed by Aedes taeniorhynchus (10.1%), Aedes albopictus (7.8%), Aedes tortillis (6.4%), Aedes aegypti (3.6%), and Cx. quinquefasciatus (3.5%). From the data in Table 4, and for some species, the number of mosquitoes collected in light traps varied significantly over the study years. For example, the mean numbers of Cx. nigripalpus per trap during 2016 and 2017 were 156.9 and 1,259.2, respectively. Figure 2 shows areas in Miami-Dade County for example, that experienced high numbers of Cx. nigripalpus during epi-week 36–38.

Table 4 Total number of adult female mosquitoes collected in BG-Sentinel traps and CDC- light traps in Miami-Dade County over 8 weeks (epi-week 33–36 and epi week 38–41) in 2016–2018.
Full size table
Fig. 2
figure 2

Source: Miami-Dade Mosquito Control and Habitat Management Division, MDC, Florida, 2018. Maps generated by the first author using ArcGIS software version 10.31

Maps of the weekly mean Cx. nigripalpus abundance in light and gravid traps from 2016 to 2018 in Miami-Dade County, Florida. 61

Full size image

Mosquito abundance pre- and post-Hurricane Irma

Figure 3 indicates that the number of mosquitoes increased immediately (one week) after Hurricane Irma at week 38 and decreased slightly beginning at weeks 39 and 40 before increasing again one week later (i.e., week 41) (Fig. 3). As indicated at 33, 34 and 35 epi-weeks in 2016, no data from the BG-Sentinel traps were reported, as these traps were established only in September 2016 (i.e., week 46). The map shows that this species was abundant post-Hurricane Irma in traps and census tracks located in the Homestead, Pembroke Pines, and Kendall neighborhoods. Most of these areas were also refuge (evacuation zones).

Fig. 3
figure 3

Source: Miami-Dade Mosquito Control and Habitat Management Division, MDC, Florida, 2018.

Estimates of mean weekly mosquito captures per trap night, 2016 to 2018, Miami-Dade County, Florida.

Full size image

Differences in mosquito abundance according to trap type pre- and post-Hurricane Irma

Table 5 presents the summary statistics for the two-way factorial ANOVA. The binary dummy variable for Hurricane Irma was a predictor of mosquito species abundance [Factorial ANOVA: F (25, 78) = 11.436, p < 0.001, R2 = 0.786] across all study census tracts in Miami-Dade County. Further, Table 5 shows that the effect of Hurricane Irma on mosquito abundance and species varied by trap type and between the two study periods (pre- and post-Hurricane Irma), suggesting that flood events can change the physical environment to favor an increase in the breeding of some mosquito species [F (1, 120) = 36.833, p < 0.001] especially for CDC light traps [F (39, 120) = 12.89; p < 0.001, R2 = 0.807] compared to pre-hurricane Irma and for mosquito species [F (19, 120) = 20.678, p < 0.001].

Table 5 Results of two-way factorial ANOVA on the effect of the hurricane variable on mosquito abundance by trap type, 2016–2018, Miami-Dade County, Florida.
Full size table

Of the 13 species of mosquitoes captured in the BG-Sentinel traps before and after Hurricane Irma, Ae. tortills (p < 0.05, 95% CI [− 0.683, − 0.139]), Cx. nigripalpus (p < 0.05, 95% CI [− 0.968, − 0.424]) and Cx. quinquefasciatus (p < 0.05, 95% CI [-0.669, -0.124]) were the most common. For CDC light traps, seven species dominated pre- and post-Hurricane Irma: Aedes tortills (p < 0.05, 95% CI [− 1.466, − 0.384]); An. atroparvus (p < 0.05, 95% CI [− 1.117, − 0.035]); Anopheles crucians (p < 0.05, 95% CI [-1.591, -0.509]; An. quadrimaculatus (p < 0.05, 95% CI [− 144, − 0.062]); Cx. erraticus (p < 0.05, 95% CI [-1.316, − 0.234]; Cx. nigripalpus (p < 0.05, 95% CI [2.400, − 1.317]; and Ps. columbiae (p < 0.05, 95% CI [− 1.334, − 0.251]) (Table 6).

Table 6 Univariate test summary results by species and trap type, 2016–2018, Miami-Dade County, Florida.
Full size table

Risk factors for mosquito abundance

The seven risk factors of interest were analyzed using OLS and GWR (Table 7). The global regression (OLS) model demonstrated that meteorological factors (rainfall and temperature during the week of mosquito collection), population density, and the percentage of the female population were negatively associated with mosquito abundance in Miami-Dade County before and after Hurricane Irma; however, the other correlations were not significant.

Table 7 Summary of the ordinary least squares (OLS) regression model, 2016–2018, Miami-Dade County, Florida.
Full size table

The GWR model showed the detailed spatial distribution of the associations between mosquito abundance and risk factors in Miami-Dade County. The correlations showed high variability in terms of geography. Unlike in the OLS model, all the independent variables were not significant (data not shown), suggesting spatial non-stationarity in the relationships between mosquito abundance and the independent variables of interest.

Discussion

This study attempted to examine adult mosquito abundance and species composition and estimate the time lags required for the development and recovery of potential vector populations before and after Hurricane Irma in Miami-Dade County, Florida. To our knowledge, this is the first study to assess the pre-post effects of hurricanes on mosquito abundance and species composition at the census tract level in Miami-Dade County. Prior studies conducted in the continental U.S. have linked the effects of storm events on mosquito abundance, but the results have been mixed45. Until recently, (since Hurricane Katrina), very few studies have monitored mosquito populations and species composition before and after storm events in the continental U.S.11,27,34,39,40,41,72. In particular, the effects of post-disaster time lags have rarely been investigated. In the context of this study, we observed a surge in mosquito populations after Hurricane Irma in Miami-Dade County. While not every mosquito is a vector, mosquito-borne arboviruses such as WNV, SLE, and EEE as well as increases in the mosquito population have been reported post storm in Florida8,46,47,48. For this reason, Florida including Miami-Dade County with their past transmission are at risk for arbovirus transmission post-storm38,46,47,48. Moreover, Miami-Dade County was the epicenter of the 2016 Zika outbreak in continental U.S.12,13,73,74,75.

To mitigate and prevent future risks, we must enhance mosquito surveillance, raise public awareness, and implement targeted vector control measures. We also found that more mosquitoes (7.3 and 8.0 times more) were captured in 2017 than at baseline (2016 and 2018). A possible explanation for this difference might be that more trap nights were carried out post-Hurricane Irma (639 trap nights) than at baseline (609 trap nights), respectively (see Table 3). We also observed an immediate increase in the mean number of mosquitoes immediately (one week) after Hurricane Irma (epi-week 38). This was followed by a slight decrease two weeks later during epi-weeks 39 and 40 before increasing again in epi-week 41 (three weeks post-Hurricane Irma). There is typically a one-week lag before mosquito activity increases due to factors like standing water accumulation and altered breeding habitats. Some research suggests this lag might extend up to four weeks, possibly due to delayed impacts on mosquito life cycles81,82. Regarding recovery, species resilience varies, with some mosquitoes rebounding faster than others. Notably, 32 species were collected in this study period and there are a diverse range of responses to effects of the hurricane therefore this is an important issue for future research. Notably, 32 species were collected during this study period, exhibiting a diverse range of responses to the effects of the hurricane. This variability underscores the importance of further research on this issue. Moise and colleagues’ study on environmental and socio-demographic predictors of the southern house mosquito, Culex quinquefasciatus, in New Orleans, Louisiana, found that highly developed areas were negatively associated with the abundance of Cx. quinquefasciatus11.

In our study, Cx. nigripalpus was by far the most common mosquito collected during the study (70.4% of all mosquito species collected). This finding is not surprising considering that this species of mosquito is the most important disease vector in Florida and is reproductively active throughout the year in South Florida83,84, particularly during the summer and early fall85. Cx. nigripalpus has been identified as a vector of SLEV86,87. In Florida, field data has also implicated this species as a predominant vector in SLEV epidemics88,89,90,91,92. Cx. nigripalpus activities are also known to increase when overall humidity increases with the onset of the rainy season84. This also accords with earlier observations, which showed that Cx. nigripalpus actively explores open areas and utilizes a range of aquatic habitats, varying in nutrient content and seasonal availability84,93. Another possible explanation for this and in the context of our study is that it’s possible that Hurricane Irma facilitated suitable conditions, particularly temperature, precipitation, and relative humidity94,95, including recently flooded habitats that allowed gravid Cx. nigripalpus females to emerge from underground culverts and gopher tortoise burrows where they overwintered, kickstarting the population surge typically seen immediately (the week) following Hurricane Irma. This finding supports evidence from previous observations84,93,96,97,98,99,100. Additionally, these findings suggest that after heavy rainfall, Cx. nigripalpus mosquitoes adapt by seeking saturated resting spots in densely vegetated areas, a finding consistent with that of Day and Curtis97 who noted Cx. nigripalpus’ daily movement as closely following rainfall patterns.

Although the responses of different mosquito species to the possible effects of Hurricane Irma varied, a note of caution is warranted. For example, mosquito surveillance conducted with CDC light traps and CDC gravid traps naturally yields different community compositions, which should be considered in future studies post Hurricanes. In our study and among the mosquito species captured in the BG-Sentinel traps, 3 of the 13 species (Ae. tortills, Cx. nigripalpus, and Cx. quinquefasciatus) were dominant post-Hurricane Irma, whereas among the mosquitoes captured using CDC light traps, 7 species were commonly collected (Ae. tortills, An. atropos, An. crucians, An. quadrimaculatus, Cx. erraticus, Cx. nigripalpus, and Ps. columbiae). Further research should also be carried out to investigate the underlying factors that contribute to this varied mosquito species response. This may further our understanding of the relationships between the amount of rainfall and other underlying factors affecting mosquito abundance.

A GWR revealed spatial non-stationarity in the relationships between mean mosquito abundance and risk factors such as human population density, rainfall, and temperature. Warmer temperatures are known to accelerate larval development, leading to the faster emergence of adult mosquitoes. However, extreme temperatures can be detrimental. Rainfall influences the availability of suitable larval habitats, with increased rain creating more breeding sites. Both temperature and rainfall are critical for mosquito survival, development, and disease transmission potential.

One of the key issues highlighted by these findings is the connection between human population density and mosquito abundance. This observation aligns with earlier research, emphasizing the intricate interplay between human communities and mosquito populations. Factors such as urbanization, population movement, and changing mobility patterns all play a role in shaping the dynamics of mosquito-borne disease transmission105,106. Overall, the results underscore the limitations of using a single set of global parameters to model the distribution of mosquito risk factors across Miami-Dade County and that temperature and rainfall during the week of mosquito collection significantly affect larval and adult survival11,101,102,103,104.

The present results are significant in at least two major respects. First, seven primary vectors of human disease predominated after Hurricane Irma in Miami-Dade County: Ae. tortilis, An. atropos, An. crucians, An. quadrimaculatus, Cx. erraticus, Cx. nigripalpus, and Ps. columbiae. Second, different environmental and sociodemographic factors influence both mosquito abundance and mosquito species in Miami-Dade County. This information can help guide those working in public health preparedness and disaster response. Additional research is needed to better understand the potential for mosquito-borne disease transmission.