Journal of Medical Entomology, XX(X), 2018, 1–11

doi: 10.1093/jme/tjy209
Population and Community Ecology Research

Aedes albopictus (Diptera: Culicidae) on an Invasive Edge:

Abundance, Spatial Distribution, and Habitat Usage of

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

Larvae and Pupae Across Urban and Socioeconomic
Environmental Gradients
Talya Shragai and Laura C. Harrington1,
Department of Entomology, Cornell University, 3131 Comstock Hall, Ithaca, NY 14853, and 1Corresponding author, e-mail: [email protected]

Subject Editor: Dr. Gabriel Hamer

Received 1 July 2018; Editorial decision 28 October 2018

Abstract
The Asian tiger mosquito [Aedes albopictus (Skuse)] is an invasive species of public health importance that is
currently expanding its range in the Northeast United States. Effective Ae. albopictus control depends on a thorough
understanding of factors influencing their abundance, spatial distribution, and habitat preference. We conducted a
series of container surveys for Ae. albopictus larvae/pupae over 2 yr across nine sites in neighborhoods along
its invasive range in southern New York. Selected sites represented a gradient of percent impervious surface and
median household income. Two hypotheses were tested: 1) Ae. albopictus larval/pupal abundance increases and
spatial distribution becomes less clustered as site-level median income decreases and percent impervious surface
increases because of increased larval habitat quality and availability; and 2)  container-level characteristics are
predictive of Ae. albopictus larval/pupal infestation across a range of sites. In 2016, neither median household
income nor impervious surface predicted Ae. albopictus abundance. In 2017, sites with greater impervious surface
were more heavily infested by some measures. In both years, Ae. albopictus larval/pupal spatial distribution
as measured by K-function was more clustered in with greater median household income. Most container
characteristics were either not predictive of Ae. albopictus or varied between years. Based on the variability of
predictive container characteristics, we conclude that identification of key containers is not useful in this region.
However, Ae. albopictus can be nonhomogenously distributed or abundant based on income level and impervious
surface. Improved control of immatures should consider these regional predictors of Ae. albopictus populations.

Keywords: container survey, urbanization, socioeconomic status, Asian tiger mosquito

In the 30 years since its introduction into Texas, the Asian tiger mos- Aedes albopictus larval and pupal abundance depends on environ-
quito [Aedes albopictus (Skuse)] has rapidly moved into the east- mental context. Impervious surface and socioeconomic status are two
ern United States. Aedes albopictus is now invading southern New important environmental measures that affect mosquito habitat qual-
York State and has become a serious health concern. The public is ity through variation in ambient temperature or availability of resting
at high risk of exposure to Ae. albopictus, because this mosquito and oviposition sites (Alirol et al. 2011, Little et al. 2017, Murdock
readily and aggressively bites humans, adults host-seek diurnally, et  al. 2017). While most studies of the related urban mosquito Ae.
and females oviposit in peridomestic habitat (Hawley 1988, Koehler aegypti agree that abundance increases with increasing impervious sur-
and Castner 1997). In addition to being a serious biting pest, Ae. face (Landau and van Leeuwen 2012, Rubio et al. 2013, De Azevedo
albopictus transmits at least 20 viral pathogens including dengue, et  al. 2018) and there is some evidence that larval habitat increases
chikungunya, West Nile, and Zika viruses, making it a potential risk with decreasing socioeconomic status (De Azevedo et al. 2018, Walker
to public health (Paupy et al. 2009a). Because most regional control et  al. 2018), this is less clear for Ae. albopictus. A  few studies have
of Ae. albopictus focuses on immature mosquitoes, it is of particular compared a limited number of sites of variable levels of urbanization,
importance to study larval and pupal ecology. To monitor and con- a measure closely related to impervious surface, but have produced
trol immatures of this species, we need a thorough understanding of conflicting results. Comparisons of Ae. albopictus density have dem-
their abundance and distribution across the region. onstrated that they can be the most abundant in suburban (Barker

Published by Oxford University Press on behalf of Entomological Society of America 2018. 1

This work is written by (a) US Government employee(s) and is in the public domain in the US.
2 Journal of Medical Entomology, 2018, Vol. XX, No. XX

et al. 2003, Tsuda et al. 2006), urban (Bagny et al. 2009, Beilhe et al. Another strategy to better target monitoring and control
2012, Li et al. 2014), and rural (Hornby et al. 1994, Braks et al. 2003a, of immature urban mosquitoes is to identify the most heavily
Honório et al. 2009) environments. In part, these studies are hard to exploited and highly productive container categories (key con-
compare because urbanization has been defined multiple ways; using a tainers) in order to focus limited resources. Although this can
more specific measure such as impervious surface can help hone in on be successful for Ae. aegypti (Maciel-de-Freitas and Lourenco-
precise aspects of the environment. From similarly conflicting studies de-Oliveira 2011), studies of Ae. albopictus have produced
of socioeconomic status, surveys of limited numbers of sites of different unclear results (Maciel-de-Freitas and Lourenco-de-Oliveira
incomes have found that adults (Unlu et al. 2011) and pupae (LaDeau 2011, Bartlett-Healy et  al. 2012, Unlu et  al. 2013). To this end,
et  al. 2013) are more abundant in sites of lower income, whereas we attempted to complement existing studies to establish what, if

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

other studies have found that larvae and adults are more abundant in any, container characteristics predict the presence of Ae. albopic-
high income sites (Becker et al. 2014) or that there was no variation tus larvae and pupae.
in density by income (Dowling et  al. 2013). A  key limitation of all To address these gaps and inconsistencies, we measured abun-
these studies is that all were conducted on a limited number of sites dance, spatial distribution, and container habitat characteristics of
of contrasting landscapes instead of those representing a full gradient Ae. albopictus larvae/pupae in sites representing full environmental
of either impervious surface or income, without which it is impossi- ranges for southern New York State (Westchester, Rockland, and
ble to analyze using a regression and therefore difficult to detect true Suffolk counties). We conducted two consecutive years of container
landscape-level patterns. surveys across the northern invasive range of Ae. albopictus to iden-
Just as it is important to study abundance by environmental tify trends in abundance and spatial patterns of larval/pupal habitat
variables, understanding the spatial distribution of larval positive across gradients of socioeconomic demographics (as measured by
containers across the landscape can serve as an important basis for household median income at the residential block level) and per-
effective action against immature life stages. The related species Ae. cent impervious surface. Although survey date range was longer in
aegypti has been shown to cluster both at the immature and adult the first year, all data were analyzed within the same date range for
stage (Getis et al. 2003). If abatement programs can identify where comparisons between the two sampling years. We tested the follow-
containers are most likely to be clustered into hotspots, those loca- ing hypotheses: 1) because increased impervious surface and lower
tions may be targeted for more efficient control. However, only one socioeconomic status can be associated with greater and more wide-
author to date has analyzed the spatial structure of Ae. albopictus spread larval habitat availability, as impervious surface increases
immature stages, identifying clustering above an expected random and as socioeconomic status decreases, Ae. albopictus larval/pupal
distribution of pupae in residential North Carolina neighborhoods abundance will increase and spatial distribution will become less
(Richards et  al. 2006, 2008). Again these studies were conducted clustered; and 2) container-level characteristics that have been pre-
in four similar neighborhoods rather than across an environmen- viously identified as important such as the type of container and
tal gradient, and it is unknown if this pattern holds in variable placement of container in shaded areas will predict infestation with
environments. Ae. albopictus larvae/pupae.

Fig. 1.  A map of the nine survey sites in Rockland, Westchester, and Suffolk counties, New York. Each shaded county indicates a surveyed county and each
number indicates a single survey site.
Journal of Medical Entomology, 2018, Vol. XX, No. XX 3

Methods container were collected. If >25 larvae were present, a subsample

of 25 was collected. All samples were collected with plastic transfer
Site Selection
pipettes and placed in a 50-ml tube with ~20 ml of container water.
Surveys were conducted in nine sites along the established range
of Ae. albopictus invasion in Rockland, Suffolk, and Westchester
2017 Surveys
counties (2017; Fig. 1). Three sites covering 0.1–0.15 km2 in each of
In the second year, three survey periods were conducted from 21
the three counties were selected in consultation with county health
July to 13 August 2017 within the peak activity period identified in
department personnel.
2016. Because the date range for 2017 differed from 2016, we only
The three counties we selected encompass a range of income lev-
compared the 2 yr for data taken within the same dates.
els (Min = $12,875, Q1 = $67,135, Q2 = $90,604, Q3 = $117,762,

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

Survey periods were conducted for 3 d each. Within each site, we
and Max  =  $250,001) and impervious surface (mean  =  31.93%,
first attempted to survey each house that was previously surveyed
SD  =  25.56). We selected nine sites within our three counties that
in 2016, with the goal of surveying at least 20 houses per site. Each
represented as full and even spread of the ranges as possible that also
house surveyed in 2016 was approached up to three times, once per
were 1) classified as residential by the county, 2) already identified as
survey period. If by the last survey period all houses had already
positive for Ae. albopictus, 3) dense enough for surveys to be feasibly
been approached three times and our goal had not been reached, we
conducted on foot, and 4) determined as safe by the county health
randomly approached new houses until we had surveyed a total of
department. Data from the 2011 National Land Cover Database
at least 20. House selection was randomized by assigning each block
(2011) were used to select residential sites that covered a range of
within a site a number, using a random number generator to select
percent impervious surface and data from the 2015 United States
one, and approaching each house on that block in sequence from
Census were consulted to select sites that covered a range of soci-
lowest address to highest address number. All other survey proce-
oeconomic strata as measured by median household income. Our
dures were the same as for 2016 (described above).
sample size of sites was comparable sizes used in other ecological
studies measuring changes across landscape gradients (Cox et  al. Mosquito Identification
2007, Grab et al. 2018).
Because of the high numbers of collected larvae/pupae and relative
time needed to identify each individual larva, we chose to rear all
Larval Surveys collected larvae/pupae to eclosion and only then classified adults to
2016 Surveys species. Different species are known to have different survival rates
Data from continuous days were defined as a survey period. Twelve in shared containers (Braks et al. 2004); to ensure that we were not
container survey periods were conducted from 28 April to 17 biasing numbers toward any one species or underestimating num-
October 2016. Surveys were conducted every other week for 3–5 d, bers, we counted all collected larvae and all identified adults from
apart from September and October, when sampling was conducted each container to calculate mortality rate. On average, mortality
once per month for 2 d. rate was 9%, and any container with >15% mortality was excluded
Sites were divided into blocks. Each house on every block was from any analysis for estimated total numbers. Mosquito larvae and
approached; if we gained permission from the homeowner, a survey pupae in 50-ml tubes were transferred within 24 h to a plastic 710-
was performed. This process was conducted until our target number ml food storage container. Collected water was supplemented with
(40 houses per site) was reached. For several sites, every block was dechlorinated tap water and one Hikari Gold Cichlid food pellet,
surveyed without reaching the target number. In this case, adjacent mean weight = 0.1781 g (Hikari, Himeji, Japan). Containers of mos-
blocks were randomly selected and added to the study site until 40 quitoes were held at 23°C until eclosion. Newly eclosed adults were
houses were surveyed. This process was randomized by assigning all removed daily with an aspirator and stored at −20 °C in plastic Petri
blocks adjacent to the study site a number and using a random num- dishes lined with filter paper. All adults from each container were
ber generator to select one. identified to species using published keys (Andreadis et al. 2005).
After obtaining permission from homeowners through door-
to-door interviews, teams of two to three trained field person- Data Analysis
nel surveyed the outdoor areas of each house. Teams identified all To enable comparisons between years, for all analyses comparing
water-holding containers suitable for larval development (as defined 2016 and 2017, data were restricted to the same date range. All data
by ability to hold at least 2 ml of water) and recorded water volume, analyses were conducted using R (R Core Team, Version 3.3.1).
water temperature, container category and material, degree of shad-
ing, presence of live and dead plant material, and estimated number
of mosquito larvae. Container category and container material were Site Metrics
each divided into commonly found types. For container category, this Median household income and percent impervious surface for each
included bucket, planter, sheeting, tarp, trash, tire, bird bath, piping/ site was calculated using ArcMap 10 GIS (ESRI, Redlands, CA) to
tubing, and toy. Container materials included plastic, metal, rubber, layer polygons of our survey sites against corresponding polygons
ceramic, stone/cement, and natural. Both category and material also generated from the American Community Survey Median Income
included an ‘other’ option for containers that did not fit into one of Block Group map (2015) and the National Land Cover Database
the above types. When a container was classified as ‘other’, the exact (2011). For median income, if a site spanned multiple block groups,
container category and/or material was specified. Degree of shading the mean median income weighted by area was calculated. For percent
was divided into fully shaded, partially shaded, and no shade and impervious surface, a 150 ml buffer was drawn around each house
was based on shading received throughout the day. The number of surveyed. This buffer size was chosen to correspond to a biologi-
larvae was scored as 0, 1, 2, 3, or 4, with scores indicating an esti- cally relevant average dispersal distance for Ae. albopictus (Niebylski
mated 0, 1–25, 26–50, 51–100, or >100 larvae, respectively. and Craig 1994, Honório et al. 2003, Marini et al. 2010, Medeiros
If larvae and/or pupae were present, the water was gently stirred et  al. 2017). The mean percent impervious surface was determined
with a plastic turkey baster. If there were <25 larvae, all larvae in the within each buffer, and then the mean of all buffers within a site was
4 Journal of Medical Entomology, 2018, Vol. XX, No. XX

calculated. To see the median household income and percent impervi- our calculated control function – the greater the positive difference
ous surface for each site, see Supp Table I (online only). between the observed distribution and the calculated control function,
the greater the degree of clustering (Supp Fig. 1C [online only]).
Site-level degree of infestation We next tested whether the degree of spatial clustering was pre-
The degree of Ae. albopictus larval/pupal infestation for each site was dicted by either percent impervious surface or household median
analyzed as total positive containers per containers surveyed, total income using a general linear model, which requires a single con-
positive containers per 100 houses surveyed, and total estimated tinuous response variable for each site. We reduced our output from
quantity of Ae. albopictus per site. This measure of total estimated the full range of spatial scales to the degree of clustering at 150 m
quantity was taken to account for container productivity. To calculate to produce a single K-value at a single scale for each site. Then, we

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

estimated quantity, we multiplied the proportion of mosquitoes from calculated the difference between the observed K-value and our gen-
a positive container sample identified as Ae. albopictus by the score erated control function at that single scale. We chose 150 m because
indicating the estimated number of total larvae in the original con- this is the established average flight range of adult Ae. albopictus
tainer (1–4). For example, if 50% of the sample was identified as Ae. and it was the most pronounced scale at which clustering occurred
albopictus and the original container was given a score of 3 (50–100 across sites.
larvae/pupae), the score for the estimated quantity of Ae. albopictus To address the degree to which estimated quantity of Ae. albop-
in the container was calculated to be 1.5. These scores were then ictus immatures were clustered within a site, we used a marked cor-
summed for each site. The scores were used rather than the absolute relation function, again following the methods of Getis et al. (2003).
numbers so as not to bias toward either end of the estimated range. This method was performed because we were measuring a continu-
A linear mixed effects model was used to predict abundance by ous variable (number of immature Ae. albopictus) rather than a cat-
median household income and percent impervious surface. Percent egorical variable (positive/negative container). Like our analysis of
positive containers, positive containers per 100 houses surveyed, and Ae. albopictus presence/absence, we generated a theoretical random
total estimated number were used as our dependent variables meas- distribution of Ae. albopictus larvae using 200 simulation envelopes
uring abundance. Site-level impervious surface, household median against which we compared observed Ae. albopictus distributions
income, year, the interaction between impervious surface and year, within the observed spatial structure of homes visited. For further
and the interaction between household median income and year explanation of the analysis, see Supp Fig. 2 (online only). From these
were used as our fixed variables. Site was included as a random distributions, clustering was again calculated at 150 m using the
effect. In the regression of percent positive containers and positive same method as for positive containers to represent the effect size of
containers per 100 houses surveyed, there was an outlying obser- clustering for each site.
vation. We conducted the analysis with and without the outlier and A multiple linear regression was performed to test if our two
reported both results. independent variables, median household income and percent imper-
vious surface, predicted our response variables, the clustering index
of positive containers, and estimated number of Ae. albopictus.
Site-level spatial distribution analysis
Spatial distribution at the site level was analyzed for both Ae. albop- Container-level analysis
ictus-positive containers and estimated number of Ae. albopictus. This analysis was carried out at the container level. A generalized lin-
All spatial analyses were performed using the spatstats and splancs ear mixed effects model using a binomial distribution with address,
packages in R (Baddeley and Turner 2005, Rowlingson et al. 2013). site, and county as nested random effects was performed to compare
To address the degree to which Ae. albopictus-positive containers our response variable – the presence or absence of Ae. albopictus –
were clustered within each of the nine sites, we used a modification to all container predictor variables (container category, container
of a bivariate K-function (Baddeley 2008), following the methods of material, water temperature, water volume, degree of shading, pres-
Getis et al (2003). This point pattern analysis assigns a binary value ence of dead and/or live plants, and sampling date). This model
(presence/absence of Ae. albopictus larvae) to each container in a site was fit using the lme4 package in R (Bates et al. 2014). We tested
and determines whether the spatial distribution of positive containers each individual predictor with a likelihood ratio test comparing the
is random, dispersed, or clustered. First, the function assigns a K-value full model to a reduced model omitting that variable. We then per-
at each spatial scale from zero to the total area of the site for the entire formed post hoc pairwise comparisons using the lsmeans package in
site (For an example of output of K-values across multiple scales for R (Lenth 2016).
a single site, see Supp Fig. 1C [online only]). A complete random dis- Water volume was highly skewed toward low-volume containers
tribution is typically then used as the control, and the greater above and was log transformed for the analysis. We confirmed that our sam-
random the observed value of K, the greater the degree of clustering. ple size was adequate to detect differences by running a power ana-
However, because homeowner availability and permission resulted in lysis for a multiple regression with six predictors, an anticipated effect
a nonrandom and incomplete distribution of houses surveyed within size of f2 of 0.15, statistical power of 0.8, and probability of 0.05.
each site area (for an example of nonrandom house distribution, see
Supp Fig. 1A [online only]), rather than use a complete random distri-
Results
bution of containers as the control, we calculated random clustering
within the spatial structure of the homes surveyed in each site. For Residential Field Sites
each site, we ran 1,000 permutations of random container distribu- For our study, median household income was ranged from $42,071
tions among the spatial structure of the houses we surveyed, using the (Yonkers) to $154,759 (Orangeburg-Tappan). Mean impervious
same number of positive containers as in our sampling data for each surface ranged from 9.12% (Orangeburg-Tappan) and 61.17%
permutation (for an example of a single permutation at one site, see (Yonkers) (see SuppMaterial [online only]). Low levels of multicol-
Supp Fig. 1B [online only]). The mean was taken of these permuta- linearity were present in our two site-level measures (variance infla-
tions and used as our control function. We then calculated the degree tion factor  =  1.47 for median household income, 1.82 for mean
of clustering at each site by comparing the observed distribution to impervious surface).
Journal of Medical Entomology, 2018, Vol. XX, No. XX 5

Table 1.  Counts of survey results by site, 2016 (A) and 2017 (B)

Site Houses Containers Larvae Aedes albopictus Percent of containers Aedes albopictus
surveyed surveyed positive positive containers Ae. albopictus positive positive containers
containers per 100 houses
surveyed

(A) 2016
Babylon 43 (23) 235 (110) 44 (22) 16 (14) 0.068 (0.127) 37.21 (60.87)
Central Islip 40 (16) 222 (95) 61 (25) 21 (14) 0.095 (0.147) 52.50 (87.50)

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

Harrison-Mamaroneck 45 (13) 190 (38) 58 (9) 9 (3) 0.047 (0.079) 20.00 (21.74)
New Rochelle 46 (26) 195 (82) 50 (31) 18 (15) 0.092 (0.183) 39.13 (57.69)
Orangeburg-Tappan 59 (37) 260 (184) 54 (39) 16 (16) 0.062 (0.087) 27.11 (43.24)
Pearl River 48 (21) 148 (67) 35 (21) 11 (6) 0.074 (0.090) 22.92 (28.57)
Smithtown 42 (24) 150 (80) 34 (18) 12 (10) 0.080 (0.125) 28.57 (41.67)
Spring Valley 45 (29) 210 (107) 62 (39) 16 (15) 0.076 (0.140) 35.56 (45.83)
Yonkers 45 (23) 210 (89) 29 (16) 13 (9) 0.062 (0.101) 28.89 (39.13)
Total 412 (212) 1820 (852) 436 (229) 132 (102) 0.073 (0.120)
(B) 2017
Babylon 29 120 42 35 0.29 120.69
Central Islip 26 141 48 42 0.30 150.00
Harrison-Mamaroneck 20 80 45 36 0.45* 175.00*
New Rochelle 27 88 26 23 0.26 88.46
Orangeburg-Tappan 25 80 27 16 0.20 64.00
Pearl River 31 87 30 20 0.23 64.52
Smithtown 28 101 29 24 0.24 85.71
Spring Valley 26 65 26 19 0.29 73.08
Yonkers 30 92 34 28 0.30 93.33
Total 242 854 316 243 0.28
*Outlying value *Outlying value

For 2016, the results for the full date range of surveys are shown first, and the results for surveys restricted to the 2017 survey dates for comparison between
the 2 yr are shown afterward in parentheses.

Numbers of Houses and Containers Surveyed surface for both measures (β  =  0.00124, P  =  0.0401; β  =  0.674,
In 2016, across the full sampling date range, we surveyed 412 houses P = 0.0412). For 2017, greater impervious surface led to higher per-
and 1,820 containers. In total, 436 containers (23.95%) had mos- cent positive containers and positive containers per 100 houses sur-
quito larvae; 132 (7.30%) were Ae. albopictus positive (Table 1A). veyed (Fig. 2A and B, Table 2). For 2016, impervious surface had no
When the results are restricted to the same period as the 2017 effect on either measure. When the outlier was retained, the percent
surveys, we surveyed 212 houses and 852 containers. About 229 of containers positive and the positive containers per 100 houses
(26.88%) of those containers were larvae/pupae positive, and 102 surveyed were not significantly predicted by site-level impervious
(12.00%) contained Ae. albopictus (Table  1A). Other species col- surface (β  =  0.00136, P  =  0.367; β  =  1.17, P  =  0.251). When Ae.
lected were Aedes japonicus, Aedes triseriatus, Culex pipiens/res- albopictus larval/pupal abundance was measured by total estimated
tuans (Diptera: Culicidae), and Anopheles punctipennis (Diptera: abundance, there was once again a significant interaction between
Culicidae). The distribution of estimated Ae. albopictus per positive year and impervious surface (β = 0.700, P = 0.0287); in 2017, sites
container was bimodal; most containers had 1–25 larvae/pupae, few of higher impervious surface had higher total estimated abundance
had 25–80 larvae/pupae, and slightly more had >80 larvae/pupae. (Fig. 2C, Table 2), but there was no effect for 2016.
In 2017, we surveyed 242 houses and 854 containers and found Median household income was not a predictor of Ae. albopic-
a much higher proportion of containers positive both for mosqui- tus larval/pupal estimated total abundance, nor was there any sig-
toes and for Ae. albopictus. About 210 of houses surveyed (82.64%) nificant interaction by year (β  =  0.145, P  =  0.621; β  =  7.18e-07,
were previously surveyed in 2016, and 32 were only surveyed in P  =  0.621). Median household income was also not predictive of
2017. About 316 containers were larvae/pupae positive (37.00%) the percent of Ae. albopictus-positive containers in each site or the
and 243 were Ae. albopictus positive (28.45%; Table  1B). Other number of containers per 100 houses surveyed and there was not a
species collected were Ae. japonicus, Ae. triseriatus, Cx. pipiens/ significant interaction by year for either measure (percent positive
restuans, Toxorhynchites rutilus (Diptera: Culicidae), and An. punc- containers: β = 0.00155, P = 0.405; β = 7.67e-07, P = 0.405 and
tipennis. The numbers of Ae. albopictus per positive container fol- positive containers per 100 houses surveyed: β = −0.145, P = 0.595;
lowed the same distribution as in 2016. β = 0.0300, P = 0.830).

Site-Level Degree of Infestation by Impervious Spatial Distribution by Impervious Surface and

Surface and Household Median Income Household Median Income
Aedes albopictus larval/pupal abundance as measured by percent We analyzed distribution of Ae. albopictus-positive aquatic habitat
positive containers and by positive containers per 100 houses sur- by median household income and impervious surface and found that
veyed had one outlying observation; when this outlier was removed, while impervious surface did not predict abundance, income did pre-
there was a significant interaction between year and impervious dict spatial distribution of Ae. albopictus larvae/pupae.
6 Journal of Medical Entomology, 2018, Vol. XX, No. XX

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

Fig. 2.  Increasing percent impervious surface predicted abundance of Ae. albopictus. In 2017, increasing percent impervious surface predicted both the number
of Ae. albopictus (A) percent positive containers, (B) positive container per 100 houses surveyed, and (C) total estimated quantity of Ae. albopictus in each site.

Distribution was significantly predicted by median income at a scale In 2016, water temperature, shading, and plants were significant
of 150 m. When the dependent variable was the degree of clustering predictors of positive containers, whereas container category, con-
of positive containers for each site, the results of the multiple regres- tainer material, and water volume were not significant predictors
sion indicated that the predictors explained 66.15% of the variance (Table 3). Aedes albopictus were more likely to be found in contain-
(R2 = 0.661, F(5, 12) = 3.94, P = 0.0132). The clustering of Ae. albopictus- ers with dead plants over no plants (coeff.  =  −1.927, SE  =  0.343,
positive containers was predicted by income with higher degree of clus- z-value  =  −5.616, P = <0.0001) and with dead plants over live
tering of positive containers and estimated number of Ae. albopictus in plants (coeff. = −2.283, SE = 0.845, z-value = −2.702, P = 0.00689).
higher income sites (β = 8.87e-8, P = 0.0177). This effect was stronger in In addition, they were more likely to be found in containers in
2016 than 2017, with a significant interaction between year and income full shade over those with no shade (coeff.  =  −2.114, SE  =  0.838,
(β = −1.073, P = 0.0366; Fig. 3A). When the dependent variable was z-value  =  −2.521, P =0.0117) or partial shade (coeff.  =  −0.649,
the degree of clustering of estimated quantity of Ae. albopictus for each SE  =  0.291, z-value  =  −2.232, P = 0.0256). Increasing water tem-
site, the predictors explained 44.14% of the variance (R2 = 0.441, F(5, perature increased the likelihood of a container being Ae. albopictus
12)
 = 5.53, P = 0.0510). There was a borderline significant interaction positive (coeff. = 0.119, SE = 0.0531, z-value = 2.248, P = 0.0246).
between year and income (β = 0.0106, P = 0.0510). For 2017, greater When number of positive containers was analyzed by container
income surface led to greater clustering of estimated quantity of Ae. characteristics for 2017, water temperature, container category,
albopictus (Fig. 3B). For 2016, income had no effect. Aedes albopictus water volume, and plants were significant predictors of Ae. albopic-
larval distribution, as measured by both positive containers (β = −6.32e- tus presence (Table 3). In contrast to 2016, two container categories
8, P = 0.346) and estimated number (β = 0.0188, P = 0.494), was not – planters and piping/tubing – significantly predicted the presence
related to impervious surface in either year. of Ae. albopictus (coeff.  =  0.871, SE  =  0.354, z-value  =  2.461,
P  =  0.0139; coeff.  =  1.0690, SE  =  0.466, z-value  =  2.293, P =
Positive Container Variables 0.0219), increasing water volume increased the likelihood of a con-
We looked for container measurements that predicted the presence tainer being positive (coeff. = 0.237, SE = 0.0476, z-value = 4.974,
of Ae. albopictus in order to identify variables that could focus P  <  0.0001), increasing water temperature decreased the likeli-
future control efforts. Together these data show that, while there hood of a container being Ae. albopictus positive (coeff. = −0.0597,
were significant variables in both years, the significant measurements SE = 0.0302, z-value = −1.976, P = 0.0481) and shading was not sig-
varied from 2016 to 2017. nificant. Similarly, as in 2016, Ae. albopictus were more likely to be
Journal of Medical Entomology, 2018, Vol. XX, No. XX 7

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

Fig. 3.  Median household income predicted greater clustering of Ae. albopictus. (A) In 2016 positive containers were more highly clustered in sites with higher
median household income and (B) in 2017 estimated quantity of Ae. albopictus were more highly clustered in sites with higher median household income.

Table 2.  Site-level predictors of Ae. albopictus abundance by three measures

Coeffcient Percent pos. Pos. containers per 100 Estimated quantity

containers houses surveyed of Ae. albopictus

Estimates (CI) P Estimates (CI) P Estimates (CI) P

Intercept 0.16 (0.09 to 0.22) 0.003 52.18 (−6.78 to 111.13) 0.192 20.44 (−2.26 to 43.13) 0.164
Year 0.11 (0.01 to 0.20) 0.113 16.96 (−11.59 to 45.50) 0.373 −2.60 (−34.69 to 29.50) 0.896
Impervious surface −0.00 (−0.00 to 0.00) 0.859 0.22 (−0.63 to 1.07) 0.683 −0.00 (−0.33 to 0.32) 0.984
Income −0.42 (−0.91 to 0.06) 0.180 −144.74 (−571.06 to 281.59) 0.595 −58.48 (−222.59 to 105.63) 0.569
Year by impervious surface 0.00124 (−0.08 to 0.002) 0.040 0.67 (0.27 to 1.08) 0.041 0.70 (0.24 to 1.16) 0.029
Year by income −0.08 (−0.77 to 0.61) 0.858 29.58 (−185.40 to 244.56) 0.830 71.69 (−160.40 to 303.78) 0.621
Random Effects
σ2 0.00 56.00 76.67
τ00 0.00 Site 461.39 Site 0.00 Site
ICC 0.08 Site 0.89 Site 0.00 Site
Observations 17 17 18
Marginal R2/conditional R2 0.906/0.913 0.579/0.954 NA

found in containers with dead plants over no plants (coeff. = −0.986, Variation Year to Year
SE = 0.217, z-value = −4.550, P < 0.0001), with dead plants over live We found the percent containers positive for Ae. albopictus
plants (coeff. = −1.875, SE = 0.590, z-value = −3.180, P = 0.00147), dramatically increased from the first to second year of this
and container material was not significant. study. Although surveys for 2016 were conducted over an entire
season of adult activity and those in 2017 were completed over
3 wk, this trend held even when comparing data from the same
Discussion period in both years. For data within the same date range for
Our study describes the abundance and distribution of Ae. albopictus both years, the proportion of containers positive for any mos-
larvae and pupae along impervious surface and socioeconomic gradi- quitoes was greater in 2017 than in 2016. However, the per-
ents on a northern border of its invasive range. Uniquely, we covered cent rise in the number of Ae. albopictus-positive containers
a full spectrum of sites along landscape gradients with unprecedented was larger than for mosquito-positive containers overall. This
scope and coverage for the Northeast region. Our first hypothesis, suggests that Ae. albopictus did not just increase because mos-
that increasing site-level impervious surface and decreasing household quito populations increased, but that they were more successful
median income would predict higher abundance and less clustered dis- than other container species. There is no obvious connection
tribution of Ae. albopictus larval habitat because of increased quality to weather; while we would expect that a hotter and wetter
and availability of larval habitat was partially supported. Our second weather would predict greater mosquito abundance, using
hypothesis that Ae. albopictus-positive larval habitat would be pre- data from the NOAA, Northeast Regional Climate Center at
dicted by container characteristics that have been previously identified Cornell University, 2017, was both slightly cooler and slightly
as important, such as shading or container type, was not consistently drier than 2016 (Supp Fig.  3 [online only]). The variation we
supported enough to be useful for targeted mosquito control. These observed over just 2 yr is not sufficient to identify long-term
results contribute to our knowledge of how landscape shapes Ae. patterns, but it highlights the necessity of systematic surveil-
albopictus distribution and abundance and identifies container charac- lance to track long-term population trends over time and more
teristics related to the presence of this species. clearly pinpoint the factors driving changes.
8 Journal of Medical Entomology, 2018, Vol. XX, No. XX

Table 3.  Container variables predictive of the presence of Ae. albopictus in 2016 and 2017

Coefficient 2016 2017

Odds ratios (CI) P Odds ratios (CI) P

Intercept 0.07 (0.01–0.30) 0.001 1.39 (0.27–7.22) 0.692

Water temp 1.08 (1.02–1.13) 0.005 0.94 (0.89–1.00) 0.048
Water volume 1.07 (0.97–1.18) 0.169 1.27 (1.16–1.39) <0.001
Container material
 Metal 0.92 (0.32–2.36) 0.875 1.11 (0.58–2.09) 0.759

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

 Rubber 4.40 (0.69–39.13) 0.135 1.50 (0.24–12.38) 0.674
 Ceramic 1.40 (0.37–4.63) 0.594 0.55 (0.19–1.55) 0.270
 Stone 1.62 (0.38–6.43) 0.495 1.37 (0.40–4.88) 0.616
 Other 0.99 (0.20–3.74) 0.991 2.56 (0.58–13.84) 0.231
Container type
 Planter 2.03 (0.77–5.19) 0.142 2.39 (1.20–4.84) 0.014
 Tarp 1.60 (0.51–4.69) 0.400 3.43 (0.83–17.54) 0.102
 Sheeting 0.76 (0.19–2.44) 0.660 0.54 (0.11–2.11) 0.403
 Tire 0.30 (0.03–2.08) 0.255 0.85 (0.10–5.25) 0.871
 Trash 0.47 (0.13–1.36) 0.197 1.41 (0.66–3.00) 0.374
 Toy 0.87 (0.35–1.99) 0.754 0.68 (0.30–1.48) 0.337
  Bird bath 0.28 (0.03–1.51) 0.180 0.63 (0.19–1.93) 0.431
 Piping/tubing 1.65 (0.32–7.00) 0.515 2.91 (1.17–7.37) 0.022
 Other 0.60 (0.28–1.23) 0.179 0.76 (0.42–1.36) 0.358
Degree of shading
  No shading 0.22 (0.06–0.62) 0.009 0.56 (0.29–1.06) 0.077
  Partial shading 0.65 (0.41–1.05) 0.077 0.87 (0.57–1.33) 0.533
Organic material
  Live plants 0.22 (0.05–0.72) 0.024 0.15 (0.04–0.46) 0.001
  Live and dead plants 1.20 (0.38–3.49) 0.745 0.61 (0.30–1.22) 0.161
  No plants 0.18 (0.10–0.32) <0.001 0.37 (0.24–0.57) <0.001
Observations 554 529
Cox and Snell’s R2/Nagelkerke’s R2 0.142/0.221 0.160/0.215

Site-Level Degree of Infestation by Impervious little research has been conducted on Ae. albopictus across its diverse
Surface and Household Median Income and wide geographic range, and our results suggest caution when
We found differences between 2016 and 2017 in the effect of imper- extrapolating results from one region to another.
vious surface on mosquito abundance; no relationship was found in We did not see a relationship between household median income
2016 and, by some measures, Ae. albopictus were more abundant and overall Ae. albopictus abundance. Socioeconomic status using
in sites of greater impervious surface in 2017. Impervious surface median income as a measure has been correlated with Ae. albopic-
is an important environmental factor for urban container breeding tus infestation. For example, Becker et al. (2014) and LaDeau et al.
mosquitoes like Ae. albopictus. Densely covered areas with concrete (2013) reported different infestation levels between sites of varied
and asphalt can create heat islands and increase the availability of socioeconomic status; however, ours is the first study conducted
habitat (Murdock et al. 2017). It is possible that sites with greater with sufficient sites along a socioeconomic gradient to have enough
impervious surface are of greater quality and can support larger Ae. statistical power to analyze income as a causative factor. Beyond
albopictus populations or sustain faster population growth. It may larval counts, one previous study found more ‘disused containers’
be that in 2016 other factors limited population size regardless of (containers not in use or regularly dumped or treated by residents)
impervious surface and abundance was not high enough to detect in low-income sites and a higher likelihood of Ae. albopictus occur-
any patterns across sites, but in 2017 better conditions allowed rence in these disused containers; concluding that lower-income
populations to grow, and they grew larger in the higher-quality, high neighborhoods are at greater risk for high Ae. albopictus abundance
impervious surface sites. However, the percent of containers posi- (Dowling et al. 2013). In our study, we did not detect differences in
tive for Ae. albopictus was only significantly predicted by impervious abundance by household median income, nor did we see any clear
surface when one outlying site was removed, and it is unclear how relationship between income and average number of suitable ovipos-
this trend would hold when analyzed over a greater number of sites. ition containers per site. Although we did not use the used/disused
This further highlights the needs for more comprehensive surveys metric of Dowling et al. (2013), we also did not find that container
over a larger range and number of sites than was within the scope category and material was a consistent predictor of Ae. albopictus
of this study. infestation. Altogether, our results reinforce that control campaigns
Our results do not support the current literature which typically should equally target sites across income groups.
portrays Ae. albopictus as a suburban or rural species, especially in
locations where their range overlaps with Ae. aegypti (Braks et al.
2003b, Rey et  al. 2006, Reiskind and Lounibos 2013). At least in
Spatial Distribution by Household Median Income
southern New York, which is free of Ae. aegypti, our results demon- and Impervious Surface
strate that Ae. albopictus populations can be higher in sites of high However, household median income did explain spatial distribution of
impervious surface. This discrepancy may stem from the fact that larvae/pupae, with Ae. albopictus-positive containers more clustered in
Journal of Medical Entomology, 2018, Vol. XX, No. XX 9

higher-income sites in 2016 and overall abundance more clustered in adult population sizes, at least for Ae. aegypti (Tun-Lin et al. 1996).
2017. These results may be connected to the distribution of the contain- This means that no conclusions can be drawn to quantify biting
ers themselves, which were more clustered in higher income sites. Ae. risk or to recommend control that targets adult populations. These
albopictus are considered poor dispersers with an average flight distance results can only be used to make recommendations regarding control
of 50–200 m (Marini et al. 2010), so even distribution of containers in targeting immature stages, such as dumping or treating containers.
low-income neighborhoods may facilitate more even dispersal of gravid In the long term, more research is necessary to address these limi-
females across the landscape. In contrast, in high-income neighborhoods, tations: a systematic, region-wide approach studying larvae, pupae,
Ae. albopictus movement may be restricted to ‘islands’ of clustered ovi- and adults will be essential to allow us to gain a clear understanding
position habitat. This is supported by a study of residential populations of Ae. albopictus ecology and population dynamics. Unfortunately,

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

in North Carolina that examined the spatial distribution of Ae. albop- current funding for county and regional programs in the Northeast
ictus-positive containers: they showed that pupal ‘hot spots’ in a subur- United States that conduct mosquito surveillance is insufficient to
ban neighborhood were aggregated around homes with large numbers perform wide-scale sampling. Most counties currently rely on lower
of potential oviposition containers (Richards et al. 2006). While control resource-input adult trapping, which is only effective over a short
campaigns should cover neighborhoods across the socioeconomic spec- range and therefore a poor tool for tracking Ae. albopictus.
trum, the specific strategy could be tailored based on income level. It General comparisons between studies conducted to date are ham-
should be noted that while our sites did represent a gradient of socioeco- pered by a lack of standardization and identifying biologically mean-
nomic level, Ae. albopictus populations thrive in US cities outside of the ingful differences among studies with methodological differences is
ranges represented here, and it is unknown how these patterns would therefore difficult. Discrepancies in resources between counties com-
hold in those locations. However, given our results, targeted control of pound this issue – adjacent counties with wide gaps in funding prod-
Ae. albopictus hot spots may be more efficient in high-income sites and uce incomparable data sets, hindering comparisons across space. We
widespread control more effective in low-income sites. cannot stress enough the importance of greater communication and
In contrast to our results for household income, impervious standardization between groups with similar objectives to ease and
surface was not a driver of spatial distribution in any year by any enhance the interpretation of collected data.
measure. Aedes albopictus vectors are notorious for their high level of plasti-
city and ability for rapid local adaptation, meaning that real ecological
Container-Level Predictors differences can exist between populations from different proximate
Even between two consecutive years of surveys within the same sites environments (Paupy et  al. 2009b). Control of this important vector
of primarily the same houses, all but one of our container charac- can be improved by tailoring efforts based on localized studies, sav-
teristic measures failed to consistently predict Ae. albopictus larvae ing time, money, and human effort. Existing mosquito surveillance and
from 2016 to 2017. While the presence of dead organic material in control strategies in the Northeast are not designed to target this species
container water positively predicted the presence of Ae. albopictus in and distribution data are inconsistent or incomplete. More standardized
both years, all the other significant factors were either unique to 1 yr studies covering a sufficiently large numbers of sites along landscape
or predictive in opposite directions between years. This is even true gradients are required to identify causative measures of ecological dif-
for measures like water volume or shading that have been linked to ferences. This is particularly important in southern New York, because
Ae. albopictus infestation in the past (Delatte et al. 2008, Unlu et al. it represents a northern boundary of Ae. albopictus expanding range.
2013). We argue that, at least for the Northeast US region, oviposition
behavior or survival postoviposition is highly dependent on environ- Supplementary Data
mental factors that vary year-to-year. As a result, focusing on certain
Supplementary data are available at Journal of Medical Entomology
containers is an inviable strategy. It may be more useful instead to
online.
identify landscape or climatological features that consistently deter-
mine Ae. albopictus population dynamics to target response.
Acknowledgments
Limitations and General Conclusions We are grateful for the assistance of Amy Isenberg (Rockland County),
Although our multiyear, multisite study is one of the largest conducted Scott Campbell and Brian Gibbons (Suffolk County), and John Ruggiero
to date for Ae. albopictus in the Northeast region, we still cannot (Westchester County) for their advice on sampling locations. We appreciate
make conclusions about long-term seasonal or subregional trends in the statistical consulting of Erika Mudrak and Patrick Sullivan and the tech-
nical assistance of Tamsin Smith, Jody Gangloff –Kaufman, the Harrington
our key drivers of interest. Larval surveys are labor intensive, limit-
laboratory group, and The Louis Calder Center. We thank Maria Diuk-Wasser,
ing the number of households that can be covered even with mul-
Courtney Murdock, Scott McArt and Katja Poveda for reviewing early drafts
tiple survey teams, and are difficult to conduct over a large region
of the manuscript. We are also grateful for our funding, in part from the
for multiple years or even months. Thorough sampling of the land- Cornell University Extension and Outreach (EOA) Fellowship awarded to
scape is difficult. Within our sites, we were unable to access all homes, TS, in part from the National Institute of Food and Agriculture (NIFA), U.S.
and it is unknown if those that gave us access were biased toward or Department of Agriculture, Hatch Project under 2017–18- 160, and in part by
against those with more or fewer containers. Although we made every the Centers for Disease Control and Prevention (U01 CK000509).
effort to locate all water-holding containers in each property, includ-
ing notoriously cryptic habitat such as gutters and corrugated drain References Cited
pipes, Ae. albopictus infest inaccessible and hidden habitat, meaning
2011. National land cover database. USGS Multi-resolution Land
we likely were not able to locate all containers. Our sample size was
Characteristics Consortium, US Department of Interior, US Geological
robust enough to meet the goals of this study despite these limitations,
Survey. https://www.mrlc.gov/contact.php (accessed 7 July 2017).
with sufficient power to detect significant factors for all measures. 2015. American community survey. United States Census Bureau.“B19013
We stress that the results of our larval survey do not necessarily Median Household Income in the Past 12 Months.”2007 – 2011 American
correlate to adult Ae. albopictus abundance; variable survivorship Community Survey. US Census Bureau’s American Community Survey
to adulthood can result in significant decoupling between larval and Office, 2013.
10 Journal of Medical Entomology, 2018, Vol. XX, No. XX

Alirol, E., L. Getaz, B. Stoll, F. Chappuis, and L. Loutan. 2011. Urbanisation State of Rio de Janeiro, Brazil. Memórias do Instituto Oswaldo Cruz 98:
and infectious diseases in a globalised world. Lancet. Infect. Dis. 11: 191–198.
131–141. Honório, N. A., M. G.  Castro, F. S.  Barros, M. D.  E. A.  Magalhães, and P.
Andreadis, T. G., M. C. Thomas, and J. J. Shepard. 2005. Identification guide C.  Sabroza. 2009. The spatial distribution of Aedes aegypti and Aedes
to the mosquitoes of Connecticut. Connecticut Agricultural Experiment albopictus in a transition zone, Rio de Janeiro, Brazil. Cad. Saude Publica.
Station, New Haven, CT. 25: 1203–1214.
Baddeley, A. 2008. Analysing spatial point patterns in R. Technical report, Hornby, J. A., D. E. Moore, and T. W. Miller, Jr. 1994. Aedes albopictus distri-
CSIRO, 2010. Version 4. https://research.csiro.au/software/r-work bution, abundance, and colonization in Lee County, Florida, and its effect
shop-notes. on Aedes aegypti. J. Am. Mosq. Control Assoc. 10: 397–402.
Baddeley, A., and R. Turner. 2005. Spatstat: an R package for analyzing spatial Koehler, P. G., and J.  Castner. 1997. Bloodsucking insects. University of

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

point patterns. J. Statist. Softw. 12: 1–42. Florida Cooperative Extension Service, Institute of Food and Agriculture
Bagny, L., H.  Delatte, N.  Elissa, S.  Quilici, and D.  Fontenille. 2009. Aedes Sciences, EDIS, Gainesville, FL.
(Diptera: Culicidae) vectors of arboviruses in Mayotte (Indian Ocean): LaDeau, S. L., P. T.  Leisnham, D.  Biehler, and D.  Bodner. 2013. Higher
distribution area and larval habitats. J. Med. Entomol. 46: 198–207. mosquito production in low-income neighborhoods of Baltimore and
Barker, C. M., S. L. Paulson, S. Cantrell, and B. S. Davis. 2003. Habitat pref- Washington, DC: understanding ecological drivers and mosquito-borne
erences and phenology of Ochlerotatus triseriatus and Aedes albopic- disease risk in temperate cities. Int. J.  Environ. Res. Public Health. 10:
tus (Diptera: Culicidae) in southwestern Virginia. J. Med. Entomol. 40: 1505–1526.
403–410. Landau, K. I., and W. J. van Leeuwen. 2012. Fine scale spatial urban land cover
Bartlett-Healy, K., I.  Unlu, P.  Obenauer, T.  Hughes, S.  Healy, T.  Crepeau, factors associated with adult mosquito abundance and risk in Tucson,
A. Farajollahi, B. Kesavaraju, D. Fonseca, G. Schoeler, et al. 2012. Larval Arizona. J. Vector Ecol. 37: 407–418.
mosquito habitat utilization and community dynamics of Aedes albopictus Lenth, R. V. 2016. Least-squares means: the R package lsmeans. J. Statist.
and Aedes japonicus (Diptera: Culicidae). J. Med. Entomol. 49: 813–824. Softw. 69: 1–33.
Bates, D., M. Maechler, B. Bolker, and S. Walker. 2014. lme4: linear mixed- Li, Y., F. Kamara, G. Zhou, S. Puthiyakunnon, C. Li, Y. Liu, Y. Zhou, L. Yao,
effects models using Eigen and S4. R package version 1.7. 1: 1–23. G. Yan, and X. G. Chen. 2014. Urbanization increases Aedes albopictus
Becker, B., P. T. Leisnham, and S. L. LaDeau. 2014. A tale of two city blocks: larval habitats and accelerates mosquito development and survivorship.
differences in immature and adult mosquito abundances between socio- Plos Negl. Trop. Dis. 8: e3301.
economically different urban blocks in Baltimore (Maryland, USA). Int. Little, E., D. Biehler, P. T. Leisnham, R. Jordan, S. Wilson, and S. L. LaDeau.
J. Environ. Res. Public Health. 11: 3256–3270. 2017. Socio-ecological mechanisms supporting high densities of Aedes
Beilhe, L. B., S. Arnoux, H. Delatte, G. Lajoie, and D. Fontenille. 2012. Spread albopictus (Diptera: Culicidae) in Baltimore, MD. J. Med. Entomol. 54:
of invasive Aedes albopictus and decline of resident Aedes aegypti in urban 1183–1192.
areas of Mayotte 2007–2010. Biological Invasions 14: 1623–1633. Maciel-de-Freitas, R., and R.  Lourenco-de-Oliveira. 2011. Does targeting
Braks, M. A., N. A. Honório, R. Lourençqo-De-Oliveira, S. A. Juliano, and L. key-containers effectively reduce Aedes aegypti population density? Trop.
P. Lounibos. 2003a. Convergent habitat segregation of Aedes aegypti and Med. Int. Health. 16: 965–973.
Aedes albopictus (Diptera: Culicidae) in southeastern Brazil and Florida. Marini, F., B. Caputo, M. Pombi, G. Tarsitani, and A. della Torre. 2010. Study
J. Med. Entomol. 40: 785–794. of Aedes albopictus dispersal in Rome, Italy, using sticky traps in mark-
Braks, M. A., N. A. Honório, R. Lourençqo-De-Oliveira, S. A. Juliano, and L. release-recapture experiments. Med. Vet. Entomol. 24: 361–368.
P. Lounibos. 2003b. Convergent habitat segregation of Aedes aegypti and Medeiros, M. C., E. C. Boothe, E. B. Roark, and G. L. Hamer. 2017. Dispersal
Aedes albopictus (Diptera: Culicidae) in southeastern Brazil and Florida. of male and female Culex quinquefasciatus and Aedes albopictus mosqui-
J. Med. Entomol. 40: 785–794. toes using stable isotope enrichment. Plos Negl. Trop. Dis. 11: e0005347.
Braks, M. A. H., N. A. Honorio, L. P. Lounibos, R. Lourenco-De-Oliveira, and Murdock, C. C., M. V.  Evans, T. D.  McClanahan, K. L.  Miazgowicz, and
S. A. Juliano. 2004. Interspecific competition between two invasive species B. Tesla. 2017. Fine-scale variation in microclimate across an urban land-
of container mosquitoes, Aedes aegypti and Aedes albopictus (Diptera: scape shapes variation in mosquito population dynamics and the potential
Culicidae), in Brazil. Ann. Entomol. Soc. Am. 97: 130–139. of Aedes albopictus to transmit arboviral disease. Plos Negl. Trop. Dis.
Cox, J., M. E.  Grillet, O. M.  Ramos, M.  Amador, and R.  Barrera. 2007. 11: e0005640.
Habitat segregation of dengue vectors along an urban environmental gra- Niebylski, M. L., and G. B. Craig, Jr. 1994. Dispersal and survival of Aedes
dient. Am. J. Trop. Med. Hyg. 76: 820–826. albopictus at a scrap tire yard in Missouri. J. Am. Mosq. Control Assoc.
De Azevedo, T. S., B. P. Bourke, R. Piovezan, and M. A. M. Sallum. 2018. The 10: 339–343.
influence of urban heat islands and socioeconomic factors on the spatial Paupy, C., H.  Delatte, L.  Bagny, V.  Corbel, and D.  Fontenille. 2009a. Aedes
distribution of Aedes aegypti larval habitats. Geospat. Health. 13: 623. albopictus, an arbovirus vector: from the darkness to the light. Microbes
Delatte, H., J. S. Dehecq, J. Thiria, C. Domerg, C. Paupy, and D. Fontenille. Infect. 11: 1177–1185.
2008. Geographic distribution and developmental sites of Aedes albop- Paupy, C., H. Delatte, L. Bagny, V. Corbel, and D. Fontenille. 2009b. Aedes
ictus (Diptera: Culicidae) during a Chikungunya epidemic event. Vector albopictus, an arbovirus vector: from the darkness to the light. Microbes
Borne Zoonotic Dis. 8: 25–34. Infect. 11: 1177–1185.
Dowling, Z., S. L. Ladeau, P. Armbruster, D. Biehler, and P. T. Leisnham. 2013. Reiskind, M. H., and L. P. Lounibos. 2013. Spatial and temporal patterns of
Socioeconomic status affects mosquito (Diptera: Culicidae) larval habitat abundance of Aedes aegypti L.  (Stegomyia aegypti) and Aedes albopic-
type availability and infestation level. J. Med. Entomol. 50: 764–772. tus (Skuse) [Stegomyia albopictus (Skuse)] in southern Florida. Med. Vet.
Getis, A., A. C. Morrison, K. Gray, and T. W. Scott. 2003. Characteristics of the Entomol. 27: 421–429.
spatial pattern of the dengue vector, Aedes aegypti, in Iquitos, Peru. Am. J. Rey, J. R., N.  Nishimura, B.  Wagner, M. A.  Braks, S. M.  O’Connell, and L.
Trop. Med. Hyg. 69: 494–505. P. Lounibos. 2006. Habitat segregation of mosquito arbovirus vectors in
Grab, H., B. Danforth, K. Poveda, and G. Loeb. 2018. Landscape simplifica- south Florida. J. Med. Entomol. 43: 1134–1141.
tion reduces classical biological control and crop yield. Ecol. Appl. 28: Richards, S. L., C. S. Apperson, S. K. Ghosh, H. M. Cheshire, and B. C. Zeichner.
348–355. 2006. Spatial analysis of Aedes albopictus (Diptera: Culicidae) oviposition
Hawley, W. A. 1988. The biology of Aedes albopictus. J. Am. Mosq. Control in suburban neighborhoods of a Piedmont community in North Carolina.
Assoc. Suppl. 1: 1–39. J. Med. Entomol. 43: 976–989.
Honório, N. A., W. D.  C.  Silva, P. J.  Leite, J. M.  Gonçalves, L. P.  Lounibos, Richards, S. L., S. K.  Ghosh, B. C.  Zeichner, and C. S.  Apperson. 2008.
and R. Lourenço-de-Oliveira. 2003. Dispersal of Aedes aegypti and Aedes Impact of source reduction on the spatial distribution of larvae
albopictus (Diptera: Culicidae) in an urban endemic dengue area in the and pupae of Aedes albopictus (Diptera: Culicidae) in suburban
Journal of Medical Entomology, 2018, Vol. XX, No. XX 11

neighborhoods of a Piedmont community in North Carolina. J. Med. Tun-Lin, W., B. H. Kay, A. Barnes, and S. Forsyth. 1996. Critical examination
Entomol. 45: 617–628. of Aedes aegypti indices: correlations with abundance. Am. J. Trop. Med.
Rowlingson, B., P. Diggle, R. Bivand, G. Petris, and S. Eglen. 2013. Splancs: Hyg. 54: 543–547.
spatial and space-time point pattern analysis. R package version: 2.01–33. Unlu, I., A. Farajollahi, S. P. Healy, T. Crepeau, K. Bartlett-Healy, E. Williges,
http://www2.uaem.mx/r-mirror/web/packages/splancs/splancs.pdf D. Strickman, G. G. Clark, R. Gaugler, and D. M. Fonseca. 2011. Area-
Rubio, A., M. V. Cardo, A. E. Carbajo, and D. Vezzani. 2013. Imperviousness as a wide management of Aedes albopictus: choice of study sites based on geo-
predictor for infestation levels of container-breeding mosquitoes in a focus of spatial characteristics, socioeconomic factors and mosquito populations.
dengue and Saint Louis encephalitis in Argentina. Acta Trop. 128: 680–685. Pest Manag. Sci. 67: 965–974.
The New York State Department of Health (ed.). 2017. NYSDOH state- Unlu, I., A. Farajollahi, D. Strickman, and D. M. Fonseca. 2013. Crouching
wide mosquito borne disease activity report. https://www.health.ny.gov/ tiger, hidden trouble: urban sources of Aedes albopictus (Diptera:

Downloaded from https://academic.oup.com/jme/advance-article-abstract/doi/10.1093/jme/tjy209/5253076 by guest on 15 February 2019

diseases/west_nile_virus/ Culicidae) refractory to source-reduction. Plos One. 8: e77999.
Tsuda, Y., W. Suwonkerd, S. Chawprom, S. Prajakwong, and M. Takagi. 2006. Walker, K. R., D. Williamson, Y. Carrière, P. A. Reyes-Castro, S. Haenchen,
Different spatial distribution of Aedes aegypti and Aedes albopictus along M. H.  Hayden, E.  Jeffrey Gutierrez, and K. C.  Ernst. 2018.
an urban-rural gradient and the relating environmental factors examined Socioeconomic and human behavioral factors associated with Aedes
in three villages in northern Thailand. J. Am. Mosq. Control Assoc. 22: aegypti (Diptera: Culicidae) immature habitat in Tucson, AZ. J. Med.
222–228. Entomol. 55: 955–963.