Paved Paradise: Belowground Parking Structures Sustain Urban Mosquito Populations in Washington, DC

Risk of vector-borne disease is of increasing concern in urban temperate regions (LaDeau et al. 2015), in part due to vector range expansion (Osland et al. 2021). In Washington, DC, for example, vectors of public health importance include but are not limited to Culex pipiens (L.) s.l., a primary vector of West Nile virus in North America (Mutebi and Savage 2009, Spielman 2010, Wu et al. 2014), and Aedes aegypti (L.), a vector of yellow fever virus, dengue virus, chikungunya, and Zika virus (Kay et al. 2000, Lima et al. 2016). Winter temperatures were thought to limit the range of Ae. aegypti to latitudes lower than 33°N and regions that remain above 10°C (Gloria-Soria et al. 2018). Culex quinquefasciatus (Say) is a member of the Cx. pipiens Assemblage commonly found between 30°N and 30°S and has similar limited tolerance for low temperatures (Samy et al. 2016), yet both species are found to persist and recur in Washington, DC (approximately 38°N) (Farajollahi et al. 2011, Lima et al. 2016). Studies speculate that subterranean habitat is used by Ae. aegypti overwintering (Lima et al. 2016, Gloria-Soria et al. 2018), and this may be true for Cx. quinquefasciatus, as well.

Although urban Washington, DC, is found at a temperate latitude, the urban heat island effect elevates daily minimum temperatures in the city above what would be found in the surrounding natural areas (Kim 1992). Furthermore, there is an abundance of subterranean human-made structures that are permeable to vector species. Such subterranean structures are naturally insulated, covered, and sometimes have temperature and humidity controls, which moderate temperature extremes. They also provide ovipositional resources like water in containers, drainage systems, sumps, and conduits, as well as vertebrate hosts for bloodfeeding (Mattingly 1963, Irwin et al. 2008, Mallis 2011, Goodman et al. 2018). This capacity for temperature moderation and provision of larval habitat may sustain vector species in temperate environments that would otherwise not support them.

In spring of 2018, we were directed to a mosquito infestation on a federal property in Southwest Washington, DC, which was proximal to 2 subterranean parking structures. The 3rd subterranean level(s) of these parking structures (approximately 11 m below grade) contained adult and juvenile mosquitoes and became our index sites, “A” and “B.” We then identified and surveyed accessible belowground structures within 1 km of these index sites for standing water, adult mosquitoes, and juvenile forms to determine the extent to which subterranean structures promote mosquito persistence in urban Washington, DC (Fig. 1). We surveyed the lowest levels of 8 unique parking structures, including the 2 index sites, and 1 dry moat surrounding a federal building (approximately 2.5 m below grade), for a total of 9 unique survey sites. We conducted our surveys over 5 dates between June and September 2018, and the index sites were always visited on the same days we surveyed a new site (Fig. 1 and Table 1). Adult mosquitoes were collected by mechanical aspiration (Insectazooka; BioQuip Products Inc., Rancho Dominguez, CA) of stairwells, walls, corners, and other potential resting sites. Standing water was measured and surveyed for mosquito eggs, larvae, and pupae visually and by dipping with a standard 13-cm, 350-ml dipper (BioQuip Products Inc.). Individuals were collected via modified transfer pipette or dipper from puddles and groundwater sumps. If no mosquitoes were visible, we dipped into the water 5–7 times before categorizing the water as “negative.”

The 2 index garages were surveyed monthly throughout the winter (December 2018 through June 2019) for standing water, but no mosquitoes were observed or collected during this time. In the summer of 2019, surveys of our index sites were repeated on a biweekly basis, but due to considerably drier conditions, no standing water or mosquito larvae were found. To determine whether gravid female mosquitoes were present in the garage even when standing water was absent, we deployed a Centers for Disease Control and Prevention gravid trap (John W. Hock Co., Gainesville, FL) at 1 index site biweekly from June through September. Adults were also collected by mechanical aspiration on 6 dates (May 15 through September 9, 2019). Live insect collections were always transported to the University of Maryland, College Park, MD. Adults were sacrificed at 20°C and morphologically identified to species or species complex. Juveniles were reared as described in Fritz et al. (2015), and emerging adults were morphologically identified and retained at −80°C.

In 2018, we found adult mosquitoes in 3 of the 9 belowground survey sites on 4 out of 5 dates (Table 1). In surveys of the other 6 sites over 3 dates that year, no standing water or mosquitoes were found (Fig. 1). Six taxa were collected from the 3 belowground structures, including 71 Cx. pipiens Assemblage, 10 Cx. salinarius Coq., 32 Ae. aegypti, 4 Ae. vexans (Meigen), 1 Anopheles sp., and 5 unidentified specimens in the genus Aedes. Notably, Ae. aegypti was collected from our subterranean sites in Southwest DC, approximately 1.5 km from Capitol Hill, where populations were previously reported (Lima et al. 2016, Gloria-Soria et al. 2018). We also collected Ae. aegypti adults aboveground and in a broader geographic area, including Foggy Bottom (4.7 km), Columbia Heights (5.3 km), Anacostia (4.0 km), and on the National Mall (2.3 km) in 2018, 2019, and 2020 (Arsenault-Benoit and Fritz, unpublished data). These observations confirm that Ae. aegypti populations are well established in Washington, DC, and their range has expanded beyond previous reports. In 2019, 39 Cx. pipiens complex and 1 Ae. albopictus (Skuse) were collected belowground via mechanical aspiration, while 266 Cx. pipiens Assemblage, 1 Cx. erraticus (Dyar and Knab), 2 Ae. vexans, 4 Ae. albopictus, and 2 Ae. aegypti adults were collected in the gravid trap.

Larvae collected in 2018 were found in puddles ranging from 0.3 to 3.5 cm deep, and from 0.5 to 3 m in diam. In one instance, 60 juveniles were collected from a puddle approximately 2 m × 1.2 m, and 0.6 cm deep at its deepest point, but much of this puddle was less than 0.3 cm deep. A single puddle yielded more than 900 individuals, collected as eggs, larvae, and pupae (Table 1). More than 1,000 larvae were collected and identified in 2018, compared with just over 100 adult specimens, suggesting that oviposition in belowground standing water was not incidental or rare. However, only members of the Cx. pipiens Assemblage, Ae. aegypti, and Ae. albopictus used these structures as ovipositional habitat. We speculate that species found as adults and juveniles seek out these structures, as opposed to species found only occasionally as adults, which may be belowground incidentally through the action of ventilation system fans or by vehicular movement, for example. Juvenile Culex and Aedes spp. shared ovipositional resources on multiple occasions, and juveniles collected from the same puddles were at different stages of development. Although standing water was observed covering variable surface areas and depths in 2018, this was rare in 2019. One severe rainstorm caused flooding at our index sites in 2019, but the flooding subsided within 3 days, and no juveniles were found. Fluctuations in the frequency and amount of precipitation impact standing water belowground and can have subsequent effects on mosquito populations.

In both years, Cx. pipiens Assemblage dominated our summer subterranean collections, but use of belowground structures by Culex spp. is not localized to Washington, DC (Mutebi and Savage 2009, Harbison et al. 2011). Instead, these habitats may provide opportunities for interbreeding among belowground molestus and aboveground pipiens and quinquefasciatus members of the Cx. pipiens Assemblage throughout their range. To assess the genetic ancestry of Cx. pipiens from our subterranean sites, we quantified the frequencies of Cx. pipiens form molestus, Cx. pipiens form pipiens, and Cx. quinquefasciatus alleles in the population. A random sample of 105 Cx. pipiens Assemblage specimens (30 from garage A, 43 from garage B in 2018, and 32 from garage A in 2019) were identified using molecular approaches. Genomic DNA was extracted with a Zymo Quick-DNA Miniprep Plus kit (Zymo Research, Irvine, CA) according to standard protocol. We first confirmed that all specimens were Cx. pipiens Assemblage members, not Cx. restuans Theobald or Cx. salinarius, using a polymerase chain reaction (PCR) by Crabtree et al. (1995). Then a pair of PCRs amplified the ACE2 and CQ11 loci of these samples to identify quinquefasciatus, pipiens, and molestus alleles following previously described protocols (Smith and Fonseca 2004, Bahnck and Fonseca 2006), but with slight modifications described in Fritz et al. (2015). We used a χ² test with a simulated P-value based on 2,000 replications of a Monte Carlo simulation to compare Cx. pipiens Assemblage allele frequencies across years, sites, and early to late season (July and September), where α = 0.05. All statistical analyses were performed in R version 3.2.6 (R Foundation for Statistical Computing, Vienna, Austria).

Genotyping revealed 29 individuals with Cx. quinquefasciatus ancestry, 30 Cx. pipiens form pipiens, 37 Cx. pipiens form molestus, and 9 individuals with unresolved identity, meaning the results of the ACE2 and CQ11 locus assays were not in agreement. We found no difference in genotype distributions between mosquitoes collected in garage A and garage B (χ² = 2.083, P = 0.397), those collected in 2018 and 2019 (χ² = 0.031, P = 1), and those collected in July or September (χ² = 0.245, P = 0.901). Consistency across the season and years suggests resident rather than transient populations. In Wuhan, China, a similar survey revealed ovipositional sites and juvenile mosquitoes, later identified as Cx. quinquefasciatus and Cx. pipiens. Rearing showed these Cx. pipiens to be both autogenous and stenogamous, indicating they may be Cx. pipiens form molestus (Wu et al. 2014). Genotyping results from the present study suggest a similar complement of Cx. pipiens Assemblage ancestries in belowground habitat, and we were able to rear a subset of the larvae we collected in culture for 3 generations without a blood meal, supporting our genetic findings that some, but not all, display the traits of Cx. pipiens form molestus.

To determine whether belowground parking structures offer refuge from extreme temperatures, we placed 4 temperature loggers (Onset Computer Corps, Bourne, MA) dispersed through the 3rd subterranean floor of parking garage A, and 4 loggers dispersed through the aboveground courtyard adjacent to the parking structure. Three of the 4 loggers placed aboveground in the courtyard were lost or removed during the study, so analysis includes data from only 1 aboveground logger. Loggers recorded temperature every 20 min between December 20, 2018, and February 27, 2019, and we redeployed the single remaining aboveground logger and belowground loggers 1 and 2 to record spring temperatures between April 1, 2019, and June 18, 2019. Subterranean temperatures never dropped below 5°C, and 17.9% of the logging period was spent below 10°C, compared with 41.8% aboveground. Aboveground temperatures also reached more extreme highs (35°C) and lows (−11°C) between December and June, whereas the temperature range belowground was narrower (5.4–29.5°C). Compared with a mean winter temperature of 5.12°C immediately aboveground, mean winter temperature belowground was 10.74°C, and belowground temperatures were significantly warmer (December 20 to February 27, t = −69.874, df = 6,588.9, P < 0.001).

To supplement our aboveground temperature data, we retrieved daily mean temperature readings for our study dates from the National Oceanic and Atmospheric Administration (NOAA) weather station at Reagan International Airport (DCA), Arlington, VA (NOAA, National Centers for Environmental Information), and calculated daily mean temperatures for each logger. A sliding window analysis of daily temperature means applied a 2-wk window and a 1-wk step using the evobiR package in R (v1.1; Blackmon and Adams 2015). Analysis of variance indicated that a linear model that included location (above- or belowground) explained significantly more of the variance in daily temperature mean than a model that excluded it (F = 25.195, df = 1, P < 0.001). This suggested that there were significant differences in the temperature changes over time between locations.

Daily temperature ranges belowground were much narrower than aboveground, and this difference in variability could have an impact on mosquito fitness and vectorial capacity (Carrington et al. 2013). Extreme fluctuations in daily temperature also are known to negatively affect Ae. aegypti survival (Lambrechts et al. 2011), and because they feed and rest indoors (Dzul-Manzanilla et al. 2017), they are well adapted to exploit urban infrastructure when available. In a study of belowground stormwater systems, Cx. quinquefasciatus egg rafts were collected belowground even when adjacent aboveground temperatures ranged from 2.8°C to 42.2°C (Harbison et al. 2011), so temperature moderation belowground may aid in survival beyond reported thermal tolerance ranges.

While we identified larvae and pupae in 2 parking structures in 2018, our 2019 results suggest they may not serve as reliable ovipositional sites every year and that an absence of mosquitoes was not due to unsuitable regional conditions, but whether the site itself provided suitable conditions. Consistent rainfall in urban areas may cause flooding, creating ovipositional habitats. The presence of adult females in both years and at multiple belowground locations highlights the value of surveying and targeting these structures for vector control. Belowground urban infrastructure may allow populations to persist for longer and build up earlier in the spring, plus warmer temperatures can shorten the development and generational time, extending the breeding season, increasing population size, and potentially affecting disease transmission. Furthermore, removal and replacement of diverse natural habitat with homogenized urban habitat skews the mosquito community toward anthropophilic species that vector human disease (LaDeau et al. 2015). An integrated pest management approach for mosquito populations should include surveillance of subterranean habitats. Removal of standing water when possible, or larvicidal treatments otherwise, may significantly contribute to reduction of mosquitoes aboveground. In years with persistent wet weather, pest control operators and local public health officials should incorporate surveillance of these structures into their programs.

The authors thank Randall Howes (GSA, retired) for assistance identifying survey locations and collecting specimens. We also thank Justin Harbison, Patrick Irwin, Paul Leisnham, and Anahí Espíndola for reading and suggesting improvements to this manuscript. This work was funded by the University of Maryland startup funds and NIH R01AI125622A to MLF.