Content uploaded by Roberto Barrera
Author content
All content in this area was uploaded by Roberto Barrera
Content may be subject to copyright.
1087
Am. J. Trop. Med. Hyg., 85(6), 2011, pp. 1087–1092
doi:10.4269/ajtmh.2011.11-0381
Copyright © 2011 by The American Society of Tropical Medicine and Hygiene
INTRODUCTION
Vector control programs could be more efficient if the spa-
tial locations of highly productive areas of Aedes aegypti were
predictable. Several studies have reported that dengue vector
abundance is highly heterogeneous, with some neighborhoods
showing significantly higher infestation levels.
1– 3 Likewise,
dengue incidence has been found to be highly variable, with a
few areas having the largest values.
4– 8 For example, most den-
gue cases (70%) were reported from a relatively small fraction
(35%) of all neighborhoods in a dengue hyperendemic city
during 6 years of observations.
4 It was observed throughout
the period of study that the rank order of dengue incidence per
neighborhood was kept practically unaltered between years.
Also, neighborhoods having the largest dengue incidence
were those neighborhoods with the highest average density of
Ae. aegypti females per house, but vector abundance was also
relatively large in neighborhoods with intermediate dengue
incidence and correspondingly low in places with low dengue
incidence.
9 A distinct feature of the neighborhoods with the
largest dengue incidence was that they also had the longest
periods of uninterrupted dengue transmission (dengue persis-
tence or endemicity). Thus, it is important to study the spatial
and temporal dynamics of dengue transmission. This knowl-
edge allows for the stratification of cities so that vector control
programs can allocate their resources more effectively.
4, 10, 11
From the point of view of operational vector control, it
would also be useful to understand the spatial and temporal
patterns of mosquito abundance at finer geographical scales,
such as city block or household levels. Getis and others
12 stud-
ied the spatial distribution of Ae. aegypti at the household level
in two neighborhoods in Iquitos, Peru. They reported that Ae.
aegypti adults clustered mostly at 10 m, with some degree of
clustering up to 30 m. Chansang and Kittayapong
13 found clus-
ters of immature Ae. aegypti up to 20 m, and Getis and oth-
ers
12 found clusters up to 10 m (households). Similarly, it has
been reported that dengue cases cluster within households.
14, 15
Studies of space–time clustering of dengue cases showed clus-
ters within and around households (< 10–15 m) and clusters
that were close in time (3–6 days).
16, 17 Thus, both Ae. aegypti
and dengue cases seem to cluster at rather short distance and
time. An important consequence of this highly clustered, local
spatial pattern is that missing some houses during vector con-
trol operations can leave intact mosquito clusters that could
repopulate the area. The primary question is whether the loca-
tion of clusters can be determined in advance for operational
vector control purposes. The household-level study of Getis
and others
12 reported that most clusters of adult Ae. aegypti did
not appear in the same places in the two surveys that they con-
ducted 3 weeks apart. Strickman and Kittayapong
18 reported
that clusters of Ae. aegypti larvae in three villages in Thailand
changed locations with the seasons. Pupal surveys conducted
at two times of the year in a southern town in Puerto Rico
showed that a significant number of households changed their
status from producers (with pupae) to non-producers (without
pupae) and vice versa between surveys.
19 Thus, it would seem
that the temporal instability of the spatial distribution of Ae.
aegypti at very fine scales precludes the localization of highly
productive premises that could be targeted for vector control.
The spatial dispersal of Ae. aegypti has been studied at the
level of city blocks. For example, Fernandes and others
2 found
clusters of immature Ae. aegypti (Breteau Index) comprising
one to three blocks in Rio de Janerio, Brazil, and they con-
cluded that analyses at the neighborhood level did not allow
for the detection of such aggregation. The size of city blocks
varies within a city and between countries, but generally, they
are around 100 m or more. Given that spatial autocorrelation
seems to fade beyond 30 m for adult Ae. aegypti and at even
shorter distances for immatures,
12 clusters of mosquitoes per
block should reflect the contributions of highly productive
households within blocks. Unfortunately, the temporal stabil-
ity or predictability of block-level clusters has not been inves-
tigated, and investigation could inform if these clusters are
useful to guide vector control operations. Investigating vector
processes at the scale of hundreds of meters may prove use-
ful. Vazquez-Prokopec and others
20 found that 95% of dengue
cases reported within the first week of onset of symptoms of
an index case occurred at less than 125 m from it during an
outbreak in Cairns, Australia.
We recently investigated the temporal dynamics of female
adults of Ae. aegypti in two neighborhoods with a history of
dengue in San Juan, Puerto Rico (CDC, unpublished). In this
study, BG-Sentinel traps (Biogents, Regensburg, Germany)
were spaced slightly over 100 m from each other to mini-
mize trap interaction that could interfere with independent
Spatial Stability of Adult Aedes aegypti Populations
Roberto Barrera *
Entomology and Ecology Activity, Dengue Branch, Centers for Disease Control and Prevention, San Juan, Puerto Rico
Abstract. Vector control programs could be more efficient by identifying the location of highly productive sites of
Aedes aegypti . This study explored if the number of female adults of Ae. aegypti in BG-Sentinel traps was clustered and
if their spatial distribution changed in time in two neighborhoods in San Juan, Puerto Rico. Traps were uniformly distrib-
uted across each neighborhood (130 m from each other), and samples were taken every 3 weeks. Global and local spatial
autocorrelations were explored. Spatial stability existed if the rank order of trap captures was kept in time. There was
lack of global autocorrelation in both neighborhoods, precluding their stratification for control purposes. Hot and cold
spots were identified, revealing the highly focal nature of Ae. aegypti . There was significant spatial stability throughout the
study in both locations. The consistency in trap productivity in time could be used to increase the effectiveness of vector
and dengue control programs.
*Address correspondence to Roberto Barrera, Entomology and
Ecology Activity, Dengue Branch, Centers for Disease Control and
Prevention, Calle Cañada, San Juan, Puerto Rico 00920. E-mail:
rbarrera@cdc.gov
1088 BARRERA
estimations of vector density. Traps were operated every
3 weeks to minimize sampling the same mosquito cohorts. We
took advantage of this setup to investigate if Ae. aegypti adults
showed spatial clustering and determine if the spatial pattern
of adult abundance within each neighborhood changed in
time. A static spatial pattern of Ae. aegypti could be valuable
for preemptive vector control measures. This study reports
significant concordance in the rank orders of trap catches
throughout the study, showing high temporal consistency or
stability in the spatial pattern of Ae. aegypti females in both
neighborhoods that may be useful for vector and dengue con-
trol operations.
MATERIALS AND METHODS
Study sites. The study was carried out in two neighborhoods
of the Metro Area of San Juan, Puerto Rico: El Comandante
(EC, 6,951 persons and 1,979 buildings; US Census 2000) and
Villa Carolina (VC, 9,240 persons and 1,996 buildings). These
two neighborhoods are 3 km apart and belong to the adja-
cent municipalities of San Juan (EC, 18°24′02″ N, 65°59′30″ W)
and Carolina (VC, 18°23′52″ N, 65°57′26″ W). Rainfall in
the San Juan area occurs year round, with a relatively short
dry season (< 100 mm/month) between January and March
and two rainy peaks around May and November. Total
annual rainfall at the nearby Muñoz–Marin International
Airport (5–7 km) during 2008 was 1,388 mm, and mean
annual temperature was 27.0°C. The population dynamics of
Ae. aegypti were investigated in these neighborhoods to even-
tually compare the impact of vector control measures (control
versus intervention).
Carolina Municipality had a spatial insecticide spraying pro-
gram (truck-mounted Ultra Low Volume [ULV] equipment)
that was active throughout the study, whereas the San Juan
Municipality used a similar insecticide spraying technique but
only around notified cases of dengue. Thus, we believe that VC
was subjected to a more frequent application of ULV insecti-
cide spraying. However, we could not establish the frequency
or coverage of insecticide spraying in either neighborhood
because of insufficient data.
Adult Ae. aegypti mosquitoes. The study consisted of
capturing Ae. aegypti adults using 40 BG-Sentinel mosquito
traps baited with BG-Lure (lactic acid, ammonia, and caproic
acid; Biogents, Regensburg, Germany) in each neighborhood
from November 2007 to December 2008 (20 samples). Each
trap was operated for 4 consecutive days every 3 weeks to
avoid collecting female Ae. aegypti from the same adult cohort
given that they are not expected to live beyond that time
in the field.
21 Collection bags were replaced every day, and
batteries were replaced after 2 days of operation. Traps were
uniformly distributed across each neighborhood, resulting
in intertrap average distances of 132 m in EC and 137 m in
VC. We calculated the average number of female Ae. aegypti
captured per trap per day for each sample as a measure of
relative abundance.
A Geographical Information System (GIS; ArcView 9.2;
Esri, Redlands, CA) was developed for each neighborhood
with the following geo-spatial layers: polygons showing house
boundaries (Tax Revenue Agency, San Juan, Puerto Rico),
points showing house centroids, lines representing streets, and
house centroids representing trap locations. Thematic maps
showing the number of Ae. aegypti females captured per trap
per day for each of the 20 samples were made to visually exam-
ine the spatial patterns over time.
The global Moran’s I correlation coefficient
22 was used to
determine if the number of Ae. aegypti females per trap per
day was spatially autocorrelated in each neighborhood. Spatial
autocorrelation occurs when the numbers of mosquitoes per
trap per day in nearby traps are more similar than in traps that
are farther away. The Moran’s I correlation coefficient is said
to be a global measure of spatial dependence, because it uses
a single summary statistics to describe the overall spatial auto-
correlation in the neighborhood. The test was applied to each
sample to determine if autocorrelation changed in time. To
detect local spatial patterns of Ae. aegypti abundance within
each neighborhood, we calculated the Getis–Ord Gi* spatial
statistics.
23 This test detects hot/cold spots or traps with unusu-
ally larger or smaller captures. Calculations were performed
using ArcView’s Spatial Analyst tool. Thematic maps showing
summaries of average trap captures and hot/cold spots were
produced for visual analyses.
Spatial stability of mosquito abundance occurs when the
observed spatial pattern repeats in time. In this case, there is
spatial stability if the order of mosquito captures per trap is
kept from one sample to the next. To test for spatial stabil-
ity, a Spearman rank correlation coefficient of trap captures
between consecutive samples (forward lag = 1) was calculated.
A significant and positive correlation coefficient indicates
that the rank order of trap yields was kept between consecu-
tive observations. Spearman’s rank correlations were also cal-
culated for all other forward time lags (2–19 lags of 3 weeks
each) to determine if the correlations faded between samples
at different future times. For example, a significant positive
correlation between samples with a forward time lag of two
means similarity in ranks of trap captures that were spaced
in time by 6 weeks. To test for overall spatial stability in each
neighborhood, we calculated a Kendall’s W coefficient of con-
cordance for each neighborhood using all 20 samples. This sta-
tistic measures the overall concordance among the rank order
of trap yields for all samples and varies between zero and one.
Significant values of Kendall’s W imply that there was overall
consistency in trap ranks throughout the study. Accumulated
rainfall during the second and third weeks before a given
sampling date was calculated to determine whether changes
in rainfall and average mosquito population were associated
with periods of spatial stability or instability. Rainfall during
the week of sampling does not contribute many new adult
Ae. aegypti , because its immature development lasts about
1 week.
RESULTS
Spatial patterns. The average number of Ae. aegypti
females per trap per day was 4.76 ± 0.22 (±95% confidence
interval [CI], N = 3,059 trap days) in EC. There was significant
spatial autocorrelation at α = 0.05 in only 1 of 20 samples
during the first week of July of 2008 ( I = 0.32, Z = 2.803).
In general, average Ae. aegypti captures in EC were spread
throughout the neighborhood, without appreciable global
clustering ( Figure 1A ). This figure depicts how traps were
spaced throughout the neighborhood, although the location
of houses, blocks, or streets is not shown. Global clustering
would typically show areas with one or more traps with large
numbers of mosquitoes surrounded by traps with numbers that
1089
SPATIAL STABILITY OF AEDES AEGYPTI
gradually decrease with the distance. Getis–Ord’s local spatial
statistics were calculated for every sample to detect traps with
unusually large (hot spots) or small (cold spots) captures. To
summarize and map hot spots throughout the study period,
we assigned a value of one to each trap that was identified
as a hot spot ( P < 0.05) in any sampling date and added the
values to represent the number of times that a trap was a hot
spot ( Figure 1B ). Likewise, traps that were classified as cold
spots ( P < 0.05) within a sampling date were assigned a value
of −1 and added up. Cold spots are traps with low values that
are surrounded by other traps with low values. Cold spots in
EC were localized close to each other, whereas hot spots were
observed throughout the neighborhood ( Figure 1B ). Despite
the lack of global spatial autocorrelation, local clustering was
common, and some traps were frequent hot spots (4–11 of 20
samples) ( Figure 1B ). It can be observed that traps with large
average captures of Ae. aegypti were usually classified as hot
spots.
The average number of Ae. aegypti females per trap per
day was 3.80 ± 0.14 ( N = 3,048) in VC. None of the Moran’s
I correlation coefficients were significant at α = 0.05 for
any sample, indicating lack of global spatial autocorrelation
throughout the study. Similar to EC, average numbers of
Ae. aegypti females per trap in VC were spread throughout
the neighborhood, which is the main reason why the global
autocorrelation analysis did not detect significant clustering
( Figure 2A ). Getis–Ord’s local spatial statistics revealed the
presence of hot and cold spots in VC ( Figure 2B ). The spatial
dispersal of hot spots corresponded well with the location of
traps with large mosquito yields. Cold spots were scattered,
particularly around the periphery of the neighborhood in VC
( Figure 2B ).
Spatial stability. The rank order of mosquito captures per
trap (every 3 weeks) was compared using the Spearman
correlation coefficient to determine the similarity of trap yields
between sampling dates. For example, the correlation between
rank orders of Ae. aegypti females per trap between samples
two and one was 0.514 ( P < 0.05) in EC. Most correlation
coefficients between consecutive samples (future time lag = 1)
were highly ( P < 0.01) significant in EC, with slight reductions
in significance ( P < 0.05) on occasions that seemed to be
associated with marked increases in rainfall and numbers of
Ae. aegypti females per trap ( Figure 3A ). Those reductions can
be observed between samples 4 and 5, 10 and 11, and 14 and
15. There were significant and negative associations between
Spearman’s rank correlations and both the number of Ae.
aegypti females per trap ( r = −0.534, P < 0.05) and rainfall
( r = −632, P < 0.01) per sampling date. That is, spatial stability
decreased at times when Ae. aegypti populations increased or
expanded because of rainfall. Average Spearman’s correlation
coefficients were largest at forward time lags one and two,
but correlations did not fade out and stayed above significant
levels (α = 0.05) for most time lags ( Figure 4 ). The Kendall’s
W coefficient of concordance of mosquito yields per trap
throughout the study was significant ( W = 0.305, P < 0.01)
in EC, showing overall consistency in the rank order of trap
yields.
Most of the Spearman’s correlation coefficients of mos-
quito captures between consecutive samples (lag = 1) in VC
were highly significant ( P < 0.01) ( Figure 3B ). Similar to EC,
Spearman’s correlation coefficients between rank orders of
trap captures were largest at forward time lags one and two
and stayed above significant levels for all time lags ( Figure 4 ).
The Kendall’s W coefficient of concordance in VC was lower
than in EC but nevertheless, significant ( W = 0.152,
P < 0.01).
There did not seem to be any consistency in the changes of
the correlation coefficients and rainfall or mosquitoes per trap
( Figure 3B ), which was observed in EC ( Figure 3A ). There was
a lack of correlation between the Spearman’s rank coefficients
and the number of Ae. aegypti females per trap ( r = 0.271,
P > 0.05) or rainfall ( r = 212, P > 0.05) in VC.
DISCUSSION
This investigation showed lack of global spatial depen-
dence of the number of female adult Ae. aegypti captured
in BG-Sentinel traps that were uniformly spaced (130 m) in
each of two neighborhoods during 20 consecutive population
samples every 3 weeks in San Juan, Puerto Rico. This finding
means that adult females of Ae. aegypti were not clustered in
particular areas of the neighborhoods, which is unfortunate;
these neighborhoods could not be stratified into areas with
varying mosquito densities that would simplify vector control
operations. This finding is possibly because of the functional
Figure 1. Spatial pattern of ( A ) the overall average number of Ae.
aegypti females per trap per day in 40 traps scattered throughout EC
neighborhood (sampled every 3 weeks; total of 20 samples) between
November 2007 and December 2008, San Juan, Puerto Rico, and
( B ) a summary of traps classified as hot or cold spots based on the
Getis–Ord’s (Gi*) statistics.
1090 BARRERA
homogeneity in terms of housing type, basic public services,
etc. of the residential neighborhoods investigated here.
The analysis of local spatial dependence did reveal local clus-
tering or hot spots scattered throughout the neighborhoods,
and the temporal analyses showed a relatively high concor-
dance in the rank order of trap productivity in time, which trans-
lates into a pattern of spatial stability of Ae. aegypti females in
both neighborhoods. Spatial stability, expressed as the persis-
tence of hot spots for periods of time at the same locations, has
been reported for tsetse flies in Luke Community, Ethiopia.
24
A previous study using BG traps revealed significant spa-
tial clustering of adult Ae. aegypti at the household scale but
little temporal clustering in individual traps that were oper-
ated for 15 days in Cairns, Australia.
25 Other previous studies
conducted at the household scale did not show spatial consis-
tency in adult or immature density in time.
12, 18, 19 This study dif-
fers from previous ones in that we sampled every 3 weeks for
over 1 year, which allows for the observation of how spatial
Figure 2. Spatial pattern of ( A ) the overall average number of Ae. aegypti females per trap per day in 40 traps scattered throughout VC neigh-
borhood (sampled every 3 weeks; total of 20 samples) between November 2007 and December 2008, San Juan, Puerto Rico, and ( B ) a summary of
traps classified as hot or cold spots based on the Getis–Ord’s (Gi*) statistics.
Figure 3. Changes in Spearman’s correlation coefficients between
consecutive sampling dates (every 3 weeks), accumulated rainfall dur-
ing the second and third weeks before mosquito sampling, and num-
ber of female Ae. aegypti per BG-Sentinel trap per day (×10) from
November 2007 to December 2008 in ( A ) EC and ( B ) VC, San Juan,
Puerto Rico. Lines were smoothed using the cubic spline function of
Excel.
Figure 4. Average Spearman’s correlation coefficients of the
number of female Ae. aegypti per BG-Sentinel trap per day between
samples in each neighborhood at various forward time lags (each
lag = 3 weeks). Significant correlation coefficients are greater than
0.313 (two-sided test, N = 40, α = 0.05). Lines were smoothed using the
cubic spline function of Excel.
1091
SPATIAL STABILITY OF AEDES AEGYPTI
patterns change in time in greater detail, and our sampling was
done at the scale of city blocks (130 m). The scale at which
observations are made seems to be an important component
that merits additional investigations. For example, Getis and
others
12 showed that the spatial dependence of Ae. aegypti
disappeared beyond 30 m in Iquitos, Peru. Exploring scale
effects can help optimize entomological surveillance and vec-
tor control.
24
The results of the present study also showed significant cor-
relations in the rank order of mosquito abundance per trap at
most forward time lags throughout the study ( Figure 4 ), which
means high predictability in the spatial pattern of Ae. aegypti
productivity. Captured mosquitoes were most likely produced
nearby, because in most mark–release–recapture studies, Ae.
aegypti adults are captured within 100 m a few days after
release,
21, 26, 27 with the exception of gravid females that can
fly longer distances in search of containers with water.
28 The
permanency of the rank orders of abundance of Ae. aegypti
females in time must reflect the existence of persistent, local
sources of mosquitoes near the traps.
29 The important conse-
quence of the existence of relative stability in the spatial pat-
tern of trap yields is that the hot spots could be targeted for a
more efficient vector and dengue control. However, this strat-
egy clearly points out that vector control organizations would
need to conduct vector surveillance at similar scales. The advent
of mosquito surveillance devices, such as the BG-Sentinel trap
or similar devices that reflect the local abundance of adult Ae.
aegypti , provides the opportunity to do this surveillance.
The spatial heterogeneity of Ae. aegypti females per trap
was considerable ( Figures 1 and 2 ). One trap captured 91
females and 153 males of Ae. aegypti in a single day in the
porch of a house. It is conceivable that, if a dengue-infected
person stays at one of such hot spots in the study areas, it could
initiate the local transmission of dengue viruses. Furthermore,
it is reasonable to propose the hypothesis that Ae. aegypti ’s
hot spots are the most likely places where dengue viruses get
established and from which dengue viruses can be exported to
other areas. It has been shown that dengue virus transmission
is highly focal in nature and associated with the abundance
of Ae. aegypti , 30 but it has not been shown if the elimination
of local hot spots could prevent the establishment of dengue
viruses. There is evidence showing that dengue infections tend
to recur at or near the same places in time,
31, 32 which might be
because of persistent Ae. aegypti ’s hot spots.
Spatial stability faded during periods of significant increases
in rainfall and high Ae. aegypti adult density in EC, which was
revealed by the negative correlations between these variables
and the Spearman’s correlation coefficients. The negative cor-
relations mean that the rank orders of trap captures drastically
changed from one sampling date to the next. This transient
change in the spatial pattern of trap captures may be indica-
tive of the recruitment of many containers that were filled with
water in the study area, but the spatial pattern in mosquito
productivity soon returned to its previous order after reduc-
tions in population density ( Figures 3 and 4 ). This observation
seems to suggest that mosquito sampling after heavy rains
would not necessarily reflect the prevalent spatial pattern of
productivity. From a vector control perspective, it implies that
vector surveillance should be conducted more frequently dur-
ing periods in which the population of Ae. aegypti expands.
Our results are strikingly similar to the results of Sciarretta
and others,
33 who have recently described patterns of spatial
stability in tsetse flies that were transiently disrupted after sig-
nificant increases in the size of the fly populations; this sta-
bility was followed by a quick return to the previous spatial
pattern associated with lower fly densities.
The effect of rain and mosquito density on the dissimilarity
of rank order trap captures was not observed in VC ( Figure 4 ).
This neighborhood was more intensely subjected to spatial
spraying of insecticides than EC, and perhaps for that reason,
Ae. aegypti adult abundance during the peak of the rainy sea-
son in VC was also smaller than in EC: (CDC, unpublished).
There is evidence that effective vector control changes the spa-
tial pattern of adult Ae. aegypti . For example, immature con-
trol measures targeting surface containers in a southern Puerto
Rican town changed the spatial pattern of adult mosquitoes
from one in which there was no clustering before control to
one in which significant clusters appeared around untreated,
underground aquatic habitats.
34 Thus, it is likely that the spa-
tial pattern of Ae. aegypti is bound to change after the applica-
tion of effective vector control measures. For this reason, it is
recommended that vector control measures be monitored for
their effectiveness in reducing adult mosquito abundance and
the spatial distribution of mosquitoes.
34
Given that hot spots tend to be stationary for periods of
time, it is likely that an approach based on adaptive popula-
tion management could result in an efficient way to reduce the
risk of local dengue transmission. Adaptive population man-
agement has been successfully applied to reduce stationary
hot spots of tsetse flies.
24 This management approach relies on
the dynamic interaction between entomological surveillance,
aimed at identifying hot spots, and application of local vector
control in and around hot spots. Clearly, adaptive management
depends on efficient vector surveillance, prompt data analy-
sis, and mapping capabilities. Future research on novel ways
to control dengue could focus on developing inexpensive but
efficient traps for adult Ae. aegypti , establishing proper scales
for trap deployment, and testing the effectiveness of adaptive
control.
Received June 17, 2011. Accepted for publication July 16, 2011.
Acknowledgments: I would like to thank Belkis Caban, Veronica
Acevedo, Manuel Amador, Andrew J. MacKay, Gilberto Felix, Juan
Medina, Angel Berrios, Jesus Flores, and Orlando Gonzalez for their
outstanding field and laboratory work and the residents of Villa
Carolina, Extension El Comandante, and El Comandante for their
support throughout the study. Interactions with vector control offi-
cials of the municipality of Carolina are much appreciated.
Author’s address: Roberto Barrera, Entomology and Ecology Activity,
Dengue Branch, Centers for Disease Control and Prevention, San
Juan, Puerto Rico, E-mail: rbarrera@cdc.gov .
REFERENCES
1. Carbajo AE , Gomez SM , Curto SI , Schweigmann NJ , 2004 . Spatio-
temporal variability in the transmission of dengue in Buenos
Aires City . Medicina (B Aires) 64: 231 – 234 .
2. Fernandes MT , Da Costa Silva W , Souza-Santos R , 2008 .
Identification of key areas for Aedes aegypti control through
geoprocessing in Nova Iguaçu, Rio de Janeiro State, Brazil . Cad
Saude Publica 24: 70 – 80 .
3. Souza-Santos R , Carvalo MS , 2000 . Spatial analysis of Aedes
aegypti larval distribution in the Ilha do Governador neighbor-
hood of Rio de Janeiro, Brazil . Cad Saude Publica 16: 31 – 42 .
4. Barrera R , Delgado N , Jiménez M , Villalobos I , Romero I , 2000 .
Stratification of a hyper endemic dengue hemorrhagic fever
city . Pan Am J Public Health 8: 225 –233 .
1092 BARRERA
5. Flauzino RF , Souza-Santos R , Oliveira RM , 2009 . Dengue,
geoprocessamento e indicadores socioeconômicos e ambien-
tais: um estudo de revisão . Rev Panam Salud Publica 25:
456 – 461 .
6. Martinez TTP , Rojas LI , Valdes LS , Remond R , 2003 . Vulnerabili-
dad espacial al dengue. Una aplicación de los sistemas de infor-
mación geográfica en el municipio Playa de Ciudad de La
Habana . Rev Cubana Salud Publica 29: 353 – 365 .
7. Mondini A , Chiaravalloti-Neto F , Sanches MGU , Lopes JCC , 2005 .
Spatial analysis of dengue transmission in a medium-sized city
in Brazil . Rev Saude Publica 39: 444 – 451 .
8. Wen TH , Lin NH , Lin CH , King CC , Su MD , 2006 . Spatial mapping
of temporal risk characteristics to improve environmental
health risk identification: a case study of a dengue epidemic in
Taiwan . Sci Total Environ 367: 631 – 640 .
9. Barrera R , Delgado N , Jiménez M , Valero S , 2002 . Eco-
epidemiological factors associated with hyperendemic dengue
haemorrhagic fever in Maracay City, Venezuela . Dengue Bull
26: 84 – 95 .
10. Gómez Dantés H , Ramos Bonifaz B , Tapia Conyer R , 1995 . El
riesgo de transmisión del dengue: un espacio para la estratifi-
cación . Salud Publica Mex 37 (Suppl): 88 – 97 .
11. Woolhouse MEJ , Dye C , Etard JF , Smith T , Charlwood JD , Garnett
GP , Hagan P , Hii JLK , Ndhlovu PD , Wuinnell RJ , Watts CH ,
Chandiwana SK , Anderson RM , 1997 . Heterogeneities in the
transmission of infectious agents: implications for the design of
control programs . Proc Natl Acad Sci USA 94: 338 – 342 .
12. Getis A , Morrison AC , Gray K , Scott TW , 2003 . Characteristics of
the spatial pattern of the dengue vector, Aedes aegypti , in
Iquitos, Peru . Am J Trop Med Hyg 69: 494 – 505 .
13. Chansang C , Kittayapong P , 2007 . Application of mosquito sam-
pling count and geospatial methods to improve dengue vector
surveillance . Am J Trop Med Hyg 76: 820 – 826 .
14. Halstead SB , Scanlon JE , Umpaivit P , Udomsakdi S , 1969 . Dengue
and Chikungunya virus infection in man in Thailand, 1962–1964
(IV. Epidemiologic studies in the Bangkok Metropolitan Area) .
Am J Trop Med Hyg 18: 997 – 1021 .
15. Waterman SH , Novak RJ , Sather GE , Bailey RE , Rios I , Gubler
DJ , 1985 . Dengue transmission in two Puerto Rican communi-
ties in 1982 . Am J Trop Med Hyg 34: 625 – 632 .
16. Morrison AC , Getis A , Santiago M , Rigau-Perez JG , Reiter P , 1998 .
Exploratory space-time analysis of reported dengue cases dur-
ing an outbreak in Florida, Puerto Rico, 1991–1992 . Am J Trop
Med Hyg 58: 287 – 298 .
17. Tran A , Deparis X , Dussart P , Morvan J , Rabarison P , Remy F ,
Polidori L , Gardon J , 2004 . Dengue spatial and temporal pat-
terns, French Guiana. 2001 . Emerg Infect Dis 10: 615 – 621 .
18. Strickman D , Kittayapong P , 2002 . Dengue and its vectors in
Thailand: introduction to the study and seasonal distribution of
Aedes larvae . Am J Trop Med Hyg 67: 247 – 259 .
19. Barrera R , Amador M , Clark GG , 2006 . Application of the Aedes
aegypti (Diptera: Culicidae) pupal survey technique in Puerto
Rico . Am J Trop Med Hyg 74: 290 – 302 .
20. Vazquez-Prokopec GM , Kitron U , Montgomery B , Horne P ,
Ritchie SA , 2010 . Quantifying the spatial dimension of dengue
virus epidemic spread within a tropical urban environment .
PLoS Negl Trop Dis 4: e920 .
21. Harrington LC , Buonaccorsi JP , Edman JD , Costero A ,
Kittayapong P , Clark GG , Sott TW , 2001 . Analysis of survival
of young and old Aedes aegypti (Diptera: Culicidae) from
Puerto Rico and Thailand . J Med Entomol 38: 537 – 547 .
22. Moran PAP , 1950 . Notes on continuous stochastic phenomena .
Biometrika 37: 17 – 33 .
23. Ord JK , Getis A , 1995 . Local spatial autocorrelation statistics: dis-
tributional issues and an application . Geogr Anal 27: 286 – 306 .
24. Sciarretta A , Girma M , Tikubet G , Belayehun L , Ballo S ,
Baumgärtner J , 2005 . Development of an adaptive tsetse popu-
lation management scheme for the Luke Community, Ethiopia .
J Med Entomol 42: 1006 – 1019 .
25. Williams CR , Long SA , Webb CE , Bitzhenner M , Geier M , Russell
RC , Ritchie SA , 2007 . Aedes aegypti population sampling using
BG-Sentinel traps in North Queensland Australia: statistical
considerations for trap deployment and sampling strategy .
J Med Entomol 44: 345 – 350 .
26. Russell RC , Webb CE , Williams CR , Ritchie SA , 2005 . Mark–
release–recapture study to measure dispersal of the mosquito
Aedes aegypti in Cairns, Queensland, Australia . Med Vet
Entomol 19: 451 – 457 .
27. Sheppard PM , Macdonald WW , Tonn RJ , Grab B , 1969 . The dynam-
ics of an adult population of Aedes aegypti in relation to dengue
haemorrhagic fever in Bangkok . J Anim Ecol 38: 661 – 702 .
28. Reiter P , Amador MA , Anderson RA , Clark GG , 1995 . Short
report: dispersal of Aedes aegypti in an urban area after blood
feeding as demonstrated by rubidium-marked eggs . Am J Trop
Med Hyg 52: 177 – 179 .
39. Maciel-de-Freitas R , Peres RC , Souza-Santos R , Lourenço-de-
Oliveira R , 2008 . Occurrence, productivity and spatial distribu-
tion of key-premises in two dengue endemic areas of Rio de
Janeiro and their role in adult Aedes aegypti spatial infestation
pattern . Trop Med Int Health 13: 1488 – 1494 .
30. Mammen MP , Pimgate C , Koenraadt CJM , Rothman AL , Aldstadt
J , Nisalak A , Jarman RG , Jones JW , Sridiatkhachorn A ,
Ypil-Butac CA , Getis A , Thammapalo S , Morrison AC , Libraty
DH , Green S , Scott TW , 2008 . Spatial and temporal clustering
of dengue virus transmission in Thai villages . PLoS Med 5:
1605 – 1616 .
31. Kittayapong P , Yoksan S , Chansang U , Chansang C , Bhumiratana
A , 2008 . Suppression of dengue transmission by application of
integrated vector control strategies at sero-positive GIS-based
foci . Am J Trop Med Hyg 78: 70 – 76 .
32. Thai KTD , Nagelkerken N , Phuong HL , Nga TTT , Giao PT , Hung
LQ , Binh TQ , Nam NV , De Vries PJ , 2010 . Geographical het-
erogeneity of dengue transmission in two villages in southern
Vietnam . Epidemiol Infect 138: 585 – 591 .
33. Sciarretta A , Tikubet G , Baumgärtner J , Girma M , Trematerra P ,
2010 . Spatial clustering and associations of two savannah tsetse
species, Glossina morsitans submorsitans and Glossina pallid-
ipes (Diptera: Glossinidae), for guiding interventions in an
adaptive cattle health management framework . Bull Entomol
Res 100: 661 – 670 .
34. Barrera R , Amador M , Diaz A Smith J , Muñoz-Jordan JL , Rosario
Y , 2008 . Unusual productivity of Aedes aegypti in septic tanks
and its implications for dengue control . Med Vet Entomol 22:
62 – 69 .