Development of a novel sticky trap for container-breeding mosquitoes and evaluation of its sampling properties to monitor urban populations of Aedes albopictus.

Abstract

Collection methods currently used for large-scale sampling of adult Stegomyia mosquitoes (Diptera: Culicidae) present several operational limitations, which constitute major drawbacks to the epidemiological surveillance of arboviruses, the evaluation of the impact of control strategies, and the surveillance of the spreading of allochthonous species into non-endemic regions. Here, we describe a new sticky trap designed to capture adult container-breeding mosquitoes and to monitor their population dynamics. We tested the sampling properties of the sticky trap in Rome, Italy, where Aedes (Stegomyia) albopictus is common. The results of our observations, and the comparison between sticky trap catches and catches made with the standard oviposition trap, are presented. The sticky trap collected significantly larger numbers of Ae. albopictus females than any other Culicidae species representing >90% of the total catches. A maximum of 83 An. albopictus females was collected in a single week. A high correlation (Pearson correlation coefficient r= 0.96) was found between the number of females and the number of eggs collected by the traps. The functional relationship between the number of eggs and the number of adult females was assessed by major axis regression fitted to log(1 +x)-transformed trap counts as y= 0.065 + 1.695x. Trap samples significantly departed from a random distribution; Taylor's power law was fitted to the trap samples to quantify the degree of aggregation in the catches, returning the equations s(2)= 2.401 m(1.325) for the sticky trap and s(2)= 13.068 m(1.441) for the ovitrap, with s(2) and m denoting the weekly catch variance and mean, respectively, indicating that eggs were significantly more aggregated than mosquitoes (P < 0.0001). Taylor's power law parameters were used to estimate the minimum number of sample units necessary to obtain sample estimates with a fixed degree of precision and sensitivity. For the range of densities encountered in our study area during the Ae. albopictus breeding season, the sticky trap was more precise and sensitive than the ovitrap. At low population densities (c. < 0.1 mosquito/trap), however, the ovitrap was more sensitive at detecting the presence of this species. Overall, our results indicate that our new model of sticky trap can be used to sample Ae. albopictus females in urban environments, and, possibly, other container-breeding Stegomyia mosquitoes (e.g. Aedes aegypti). The technical properties of the new trap are discussed with respect to its possible application in monitoring the population dynamics of container-breeding mosquitoes, in studying their bionomics, and in vector surveillance and, possibly, control.

Full text

Page 1

Medical and Veterinary Entomology (2007), 21, 183–195
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society 183
Development of a novel sticky trap for container-
breeding mosquitoes and evaluation of its
sampling properties to monitor urban
populations of Aedes albopictus
L. FACCHINELLI 1 , L. VALERIO 1 , M. POMBI 1 , P. REITER 2 ,
C. COSTANTINI 3,4 and A. DELLA TORRE 1
1 Parasitology Unit, Department of Public Health, University of Rome ‘ La Sapienza ’ , Rome, Italy, 2 Unit of Insects and Infectious
Diseases, Pasteur Institute (Institut Pasteur), Paris, France, 3 Research Unit #016 ‘ Population Biology and Control of Insect Disease
Vectors ’ , Institute of Research for Development (Institut de Recherche pour le Développement [IRD]), Bobo-Dioulasso and
4 Institute of Research in Health Sciences (Institut de Recherche en Sciences de la Santé), Bobo-Dioulasso, Burkina Faso
Abstract . Collection methods currently used for large-scale sampling of adult
Stegomyia mosquitoes (Diptera: Culicidae) present several operational limitations,
which constitute major drawbacks to the epidemiological surveillance of arboviruses,
the evaluation of the impact of control strategies, and the surveillance of the spreading
of allochthonous species into non-endemic regions. Here, we describe a new sticky trap
designed to capture adult container-breeding mosquitoes and to monitor their popula-
tion dynamics. We tested the sampling properties of the sticky trap in Rome, Italy, where
Aedes (Stegomyia) albopictus is common. The results of our observations, and the com-
parison between sticky trap catches and catches made with the standard oviposition trap,
are presented. The sticky trap collected significantly larger numbers of Ae. albopictus
females than any other Culicidae species representing >90% of the total catches. A maxi-
mum of 83 An. albopictus females was collected in a single week. A high correlation
(Pearson correlation coefficient r = 0.96) was found between the number of females
and the number of eggs collected by the traps. The functional relationship between the
number of eggs and the number of adult females was assessed by major axis regression
fitted to log(1 + x )-transformed trap counts as y = 0.065 + 1.695 x . Trap samples sig-
nificantly departed from a random distribution; Taylor ’ s power law was fitted to the trap
samples to quantify the degree of aggregation in the catches, returning the equations
s 2 = 2.401 m 1.325 for the sticky trap and s 2 = 13.068 m 1.441 for the ovitrap, with s 2 and
m denoting the weekly catch variance and mean, respectively, indicating that eggs were
significantly more aggregated than mosquitoes ( P < 0.0001). Taylor ’ s power law
parameters were used to estimate the minimum number of sample units necessary to
obtain sample estimates with a fixed degree of precision and sensitivity. For the range of
densities encountered in our study area during the Ae. albopictus breeding season, the
sticky trap was more precise and sensitive than the ovitrap. At low population densities
( c. < 0.1 mosquito/trap), however, the ovitrap was more sensitive at detecting the pres-
ence of this species. Overall, our results indicate that our new model of sticky trap can
Correspondence: Dr Alessandra Della Torre, Parasitology Unit, Department of Public Health, University of Rome ‘ La Sapienza ’ , Piazzale Aldo
Moro 5, 00185 Rome, Italy. Tel.: +39 06 4991 4932/4933; Fax: +39 06 4991 4653; E-mail: ale.dellatorre@uniroma1.it , for trap methodology;
Dr Carlo Costantini, Institut de Recherche pour le Développement (IRD), 01 BP 171, Bobo Dioulasso, Burkina Faso. Tel.: +226 2097 1269/4987;
Fax: +226 2097 0942; E-mail: carlo.costantini@ird.bf , for statistical analyses

Page 2

184 L. Facchinelli et al.
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
Introduction
Container-breeding mosquitoes of the subgenus Stegomyia , pri-
marily Aedes aegypti (L) and Aedes albopictus (Skuse) (Diptera:
Culicidae), represent a major threat to health in the tropics as
they are the most efficient vectors of significant arboviruses
such as yellow fever, dengue and Chikungunya. Moreover, they
are significant nuisance pests wherever their distribution ex-
tends to temperate regions, as in the case of Ae. albopictus in
North America and Europe ( Gratz, 2004 ).
Adult mosquito collections provide essential samples for dis-
ease surveillance and vector monitoring. Information that is
most relevant at the epidemiological level can be gathered by
collecting host-seeking females. For a long time, the gold stand-
ard for this purpose has involved the collection of mosquitoes
that land on human volunteers, but this technique is nowadays
considered unethical whenever there is an unacceptable likeli-
hood of exposing the volunteers to the risk of contracting non-
preventable and potentially life-threatening diseases. To date,
light traps have been widely used to catch crepuscular and noc-
turnal host-seeking mosquitoes, but they are of little use for day-
flying Stegomyia ( Service, 1993 ). Collections of host-seeking
females of this subgenus are therefore mainly performed using
traps baited with host-related semiochemicals or visual stimuli
(e.g. Fay − Prince, CDC Wilton or BG-Sentinel traps), which
provide reliable samples, but which, in general, are not cost-
effective and require a source of power. Alternatively, active
collection techniques such as backpack aspirators ( Clark et al. ,
1994; Scott & Morrison, 2003 ) can be employed to collect all
fractions of the adult mosquito population, thus providing indi-
rect indices of mosquito abundance and vector − host contact.
Although aspirators are extensively used and catch large num-
bers of Ae. aegypti , they are labour-intensive and require a
source of power. Moreover, indices derived from manual collec-
tions can be biased by collector efficiency, site of collection (e.g.
indoors vs. outdoors), house size, the presence of furniture and
duration of sampling.
The absence of an efficient and sensitive collection method
for large-scale sampling of adult container-breeding Stegomyia
represents a major drawback to the epidemiological surveil-
lance of arboviruses, as well as the evaluation of the impact of
control strategies, and the surveillance of the spread of alloch-
thonous mosquitoes into non-endemic regions ( Scott &
Morrison, 2003 ). The development of new operational tech-
niques to collect adult females of container-breeding Stegomyia ,
and to monitor their densities, is therefore considered a most
valuable contribution to the prevention and control of arbovi-
ruses, such as dengue ( Scott & Morrison, 2003 ).
Ovitraps represent a sampling technique that is widely used
to gather indirect indices of adult abundance of container-breeding
species, derived from the number of eggs laid, and to assess
their spatial/temporal distribution. Ovitraps are black jars filled
with water and provided with a hardboard paddle on which
Stegomyia females lay eggs ( Fay & Eliason, 1966; Service,
1993 ). The great popularity of ovitraps is attributable to two
crucial properties: they are inexpensive and they are simple to
assemble and operate. For example, ovitraps have been used to:
(a) investigate the ecological parameters of Ae. aegypti and Ae.
albopictus in relation to both eco-climatic factors ( Mercado
Hernandez et al. , 2006 ) and dengue surveillance ( Chen et al. ,
2005 ); (b) study the dispersal of dengue vectors ( Liew & Curtis,
2005 ), and (c) evaluate the efficacy of vector control strategies
( Vezzani et al. , 2004 ). Although ovitrap collections are of sig-
nificant operational value, they represent a poor proxy for meas-
uring adult abundance because Ae. aegypti and Ae. albopictus
females scatter the eggs produced during each gonotrophic
cycle in many different breeding sites ( Rozeboom et al. , 1973;
Hawley, 1988; Clements, 1999 ).
The above limitation may be overcome by the use of sticky
ovitraps, which are small, dark-coloured jars similar to ovitraps,
supplied with adhesive strips on which the mosquitoes become
stuck when they land on the trap ’ s internal face. Sticky ovitraps
allow a direct count of the actual number of females visiting the
trap. Moreover, when used in areas of sympatry of different spe-
cies of the Stegomyia subgenus, whose eggs are morphologi-
cally indistinguishable, adults collected by sticky ovitrap are
easily identified, whereas ovitraps require that eggs are left to
hatch and larvae reared to the fourth instar or adult stage, thus
delaying identification for a week or longer ( Ritchie et al. ,
2003 ). Different designs of sticky ovitraps have been developed
and applied in the field. Ordonez Gonzalez et al. (2001 ) tested
the efficacy of a sticky ovitrap lined with an adhesive paper strip
in mark − release − recapture (MRR) experiments in a dengue-
endemic area of Guadelupe, Mexico. Ritchie et al. (2003, 2004)
tested a similar type of sticky ovitrap supplemented with ovipo-
sitional attractants in an Ae. aegypti- infested area in Cairns,
Australia, and showed an association between the number of
adult females captured and the risk of dengue transmission.
Sticky ovitraps have also been used for surveillance and behav-
ioural investigations of Ae. aegypti and Aedes polynesiensis in
Moorea, French Polynesia ( Russell & Ritchie, 2004 ) and in the
Torres Straits, Australia ( Ritchie et al. , 2006 ).
be used to sample Ae. albopictus females in urban environments, and, possibly, other
container-breeding Stegomyia mosquitoes (e.g. Aedes aegypti ). The technical properties
of the new trap are discussed with respect to its possible application in monitoring the
population dynamics of container-breeding mosquitoes, in studying their bionomics,
and in vector surveillance and, possibly, control.
Key words . Aedes , Stegomyia , container-breeding mosquitoes , monitoring , optimal
sampling plans , ovitrap , sticky trap .

Page 3

Development and evaluation of a sticky trap for container-breeding mosquitoes 185
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
In this paper, we describe a novel type of sticky ovitrap (here-
after referred to as ‘ sticky trap ’ ) we have developed (patent
pending) for container-breeding mosquitoes. Unlike other types
of sticky ovitrap, which are coated with adhesive material only
on the inner surface of the water container, just above the water
interface, our trap is designed to maximize the size of the sticky
area, and hence increase the chances of catching mosquitoes
when they land on the internal walls. We describe here the sam-
pling properties of the new sticky trap, based on collections made
in two study areas in Rome, Italy, where Ae. albopictus is abun-
dant, and report the results of a 2-year trial conducted to evaluate
the efficacy of the new sticky trap for monitoring the population
dynamics of Ae. albopictus in comparison with standard ovitrap
collections. The results of these trials are discussed in relation to
the use of this novel tool for monitoring the population dynamics
of container-breeding mosquitoes, for studying their bionomics
and for disease and vector surveillance.
Materials and methods
Sticky trap description
Our prototype sticky trap ( Fig. 1a ) is made of black acryloni-
trile butadiene styrene (ABS) and is composed of two elements.
The first is a water-holding container, moulded in the shape of
an inverted, truncated cone (diameter at base 8.5 cm, diameter at
top 12 cm, height 13.5 cm; Fig. 1b ). The container has four gul-
lies moulded into its sides to hold the second element, which
consists of two panels that intersect perpendicularly (height
19 cm) and which subdivide the upper volume of the trap into
four quarters. A round top (diameter 13.5 cm) lies on the two
panels, entirely covering the container to provide shelter from
sunshine and rain (inverted in Fig. 1c ). The walls of the panels in
each of the four quarters are lined with transparent plastic sheets
(overhead transparency sheets) coated with glue used for rodent
and insect control (Zapicol; Zapi Industrie Chimiche SpA,
Conselve, Padova, Italy) ( Fig. 1d ). The glue is formulated to tol-
erate long periods of heat and humidity while maintaining its
adhesive properties. Four small pieces of white plastic (1 × 3
cm), attached to the edges of the sections near the top, hold the
coated sheets in place ( Fig. 1a, c ). Similarly, a round, white plas-
tic base (diameter 16.5 cm) is attached to the base of the trap to
increase its stability. The contrast between the black body of the
trap and the white base and sheet holders provides a strong vis-
ual stimulus to approaching mosquitoes.
Study sites and field sampling
We tested the sampling properties of our new sticky trap in
two study areas in Rome. Site 1 was the internal courtyard
(about 0.4 ha) of a building located in a residential area. Eight
sticky traps (20 traps/ha) were serviced daily from 6 September
to 7 October 2003 and on a 10-day schedule from 8 August to
6 September and from 7 October to 6 December 2003. Site 2
was the Zoological Gardens ( Bioparco di Roma ), a 16-ha
area located inside a large urban park about 1.5 km from site 1,
characterized by high levels of Ae. albopictus infestation
( Di Luca et al. , 2001; Pombi et al. , 2003 ). Sixty sticky traps
were deployed throughout the area (3.75 traps/ha) from 24 July
2003 to 30 June 2004 (phase 1: weeks 1 – 49). From 1 July 2004
to 29 July 2005 (phase 2: weeks 50 – 105), the number of traps
was reduced to 20 (1.25 traps/ha). The population dynamics of
Ae. albopictus in the area were monitored in parallel for the
whole 2-year period by means of weekly collections with 20
ovitraps. During the winter months (weeks 21 – 37 and 73 – 89),
both traps were serviced every 2 – 4 weeks. Both types of trap
were placed in shaded sites, but were sufficiently exposed to be
readily visible.
Sticky traps were serviced as follows: (a) the sticky sheets
were removed and mosquitoes were morphologically identified
and counted by species and gender immediately in the field; (b)
the water container was emptied and filled with approximately
400 mL fresh tap water, and (c) the trap was equipped with new
transparent sheets freshly coated with glue. During daily collec-
tions at site 1, we marked the positions of stuck mosquitoes on
the backs of the adhesive sheets with permanent markers, using
a different colour for each day, and renewed the sheets weekly.
Ovitraps were constructed and serviced according to standard
practices ( Di Luca et al. , 2001; Reiter & Nathan, 2001 ). Small,
black plastic pots filled with 400 mL tap water were provided
with a masonite paddle (20 × 2 cm) as an oviposition substrate.
Ovitraps were serviced at the same time as the sticky traps: the
masonite paddle was removed and the ovitrap was emptied,
scrubbed and provided with a new clean paddle. Fresh tap water
was then poured into the container and the ovitrap was put back
in place. Removed paddles were placed in single plastic bags
and brought to the laboratory, where the number of eggs was
counted under a dissecting microscope. To verify that Ae. al-
bopictus was the only mosquito in our study area that laid eggs
on the paddles, some eggs were kept and allowed to hatch so
that the larvae could be used to identify samples to species.
Aedes albopictus was the sole species identified in these tests.
Traps that were found without water, or that were lost during the
trial, were replaced the following week and were discarded from
data analysis.
At site 2, a preliminary MRR experiment was carried out in
August 2003: 62 female and 20 male Ae. albopictus were col-
lected by human landing catches, marked with fluorescent pig-
ments (Day-Glo Color Corp., Cleveland, OH, U.S.A.) and
released in the centre of the study area. The sticky sheets that
were removed during the weekly service were brought to the
laboratory for the following 2 weeks, observed under ultraviolet
(UV) light and fluorescent mosquitoes recorded.
Statistical inference
Collection data are presented as the total number of mosqui-
toes caught and the geometric mean number of Ae. albopictus
females caught over 10 days (site 1) or each week (site 2). Count
data from trapping devices generally exhibit several undesirable
properties for the estimation of statistical parameters, of which
non-normal distribution and non-constancy of variance repre-
sent the most serious departures from the assumptions of

Page 4

186 L. Facchinelli et al.
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195

Fig. 1. Photographs showing details of the new sticky trap developed and tested in this study. (a) The trap assembled as operative. (b) Close-up of the
water-holding container. (c) The trap ’ s disassembled top and intersecting sections. (d) Close-up of the adhesive plastic sheets.

Page 5

Development and evaluation of a sticky trap for container-breeding mosquitoes 187
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
parametric statistical models. Transformation of trap counts can
alleviate the problem of heteroscedasticity in the data. In our
case, we assumed a log-normal distribution and transformed
trap catches to log(1 + x ) prior to analysis; the addition of 1 to
the counts prior to transformation was necessary as a result of
the frequent presence of traps containing no eggs or mosquitoes
(i.e. zero counts). Then, 10-day and weekly mean catches were
calculated from the transformed data. Subsequently, mean
catches were back-transformed to a linear scale to give weekly
geometric mean catches; these are usually referred to as
Williams ’ means.
One of our aims was to investigate whether the sticky trap
provided samples that were consistent with the dynamics of the
population of Ae. albopictus present in our study area, as in-
ferred from an independent and proven sampling technique (i.e.
the ovitrap). Hence, we aimed to verify whether the two sam-
pling techniques gave correlated catches, and assessed the func-
tional relationship of the weekly mean catch of an ovitrap with
that of a sticky trap. The degree of correlation between catches
of Ae. albopictus collected by sticky traps and ovitraps was
evaluated by calculating the Pearson product − moment correla-
tion coefficient. Homogeneity of the correlation coefficient be-
tween the two sampling phases (which corresponded to two
slightly different sampling plans [cf. above]) was tested by cal-
culating a chi-square statistic according to the procedure de-
scribed in Sokal & Rohlf (1995) . As both the sticky trap and
ovitrap were subject to sampling error, the functional relation-
ship between these sampling techniques could not be assessed
by ordinary regression. We fitted a type 2, or major axis, regres-
sion line instead ( Sokal & Rohlf, 1995 ).
Sampling properties of trapping devices, which are evaluated
to determine optimal sampling plans, depend on the spatial dis-
persion of the sampled population. We assessed the dispersion
parameters of Ae. albopictus using our weekly collections at site
2 by individual sticky traps and ovitraps as sample units, by
calculating the variance: mean ratio (VMR) of weekly catches.
In the case of a population that is dispersed at random, the VMR =
1; populations regularly distributed in space are characterized
by values of VMR < 1, whereas values of VMR > 1 indicate an
aggregated distribution. We assessed whether the VMR departed
significantly from unity by applying the chi-square test accord-
ing to Elliott (1977 ; pp. 40 – 44). An alternative approach is to
assume that trap catches are aggregated, in which case one
needs to estimate the degree of aggregation present in the sam-
ples. A popular statistical model used to describe the spatial dis-
persion of an aggregated population is the negative binomial
distribution. This is characterized by the parameter k , which can
be used as an index of aggregation. In a series of samples, a
common k ( k c ) can be approximated from the slope of the re-
gression line relating x ’ = m 2 – s 2 / n and y ’ = s 2 – m , where m is
the mean and s 2 the variance of samples of size n ( Elliott, 1977 ;
pp. 63 – 66). The regression line passes through the origin and
has slope 1/ k c . This estimation is possible only when there is no
relationship between m and 1/ k c .
A more general relationship used to quantify the degree of
aggregation in samples is based on Taylor ’ s power law. This
statistical model is an empirical functional relationship that lin-
early relates on a logarithmic scale the variance ( s 2 ) and the
arithmetic mean ( m ) of samples such that s 2 = am b , where b is
an index of aggregation, and a is a constant dependent on the
size of the sample unit and environmental conditions ( Taylor,
1961 ). The value of b is independent of mean density and many
studies have shown that it is usually a constant characteristic of
a species in a particular environment. Uniform, random and ag-
gregated distributions are, respectively, expressed by values of
b < 1, b = 1 and b > 1 ( Taylor, 1984; Kuno, 1991 ).
We followed Pitcairn et al. (1994) , who elaborated two ap-
proaches to optimal sampling plans. On one hand, the quantita-
tive analysis of dispersion is useful in determining the pattern
and number of samples necessary to obtain mean estimates with
a required level of precision. For Taylor ‘ s power law model, the
minimum number N of samples to be collected to obtain sample
estimates of mean density of fixed proportion ± d (hereafter,
sample precision ) is:

2
2
/2
b
a
d
Z
Na m
⋅
−






=

(1)
where Z is the standard normal distribution value for a given
probability ? ( Buntin, 1994 ). By contrast, when the density of a
population is low, it can be useful to know what sample size N *
is necessary to obtain at least one individual with a given prob-
ability of success (hereafter, sample sensitivity [ Pitcairn et al. ,
1994] ). This is given by:

0
log (1
log
)
e
e
N
P
α
∗
−
=

(2)
where P 0 is the probability of collecting no individuals; using
Taylor ’ s power law, P 0 can be estimated as:

P
amm
m
b
m
b
am m
0
1
2
=+
−






−
−

(3)
Results
Sticky trap catches
We observed that mosquitoes that landed on the adhesive sur-
face initially remained erect, with only the legs and/or wings
attached to the glue, thereby retaining their morpho-taxonomical
character in good view and shape. Later, however, mosquitoes
fell laterally or collapsed ventrally, and became completely
stuck to the adhesive sheets. Despite this, most of the specimens
retained visible diagnostic characters even after 10 days, and
only a few could not be identified, as reported by Ritchie et al.
(2003) . Although we did not systematically record the gono-
trophic stage, we observed that mainly gravid females were col-
lected; however, unfed and freshly fed females were also found.
In some cases, eggs extruding from the abdomens of gravid fe-
males were observed attached to the sticky surface. Larvae were
found in the sticky traps only very rarely. Geckos and lizards
were occasionally found stuck to the adhesive sheets and snails
were observed resting in the traps.

Page 6

188 L. Facchinelli et al.
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
Site 1 . By the end of the 4-month sampling period, the sticky
trap had collected 1606 Ae. albopictus in 958 trap days. Most of
these (85.7%) were females, thus resulting in a mean collection
rate of 1.4 females/trap/day. The maximum geometric mean
number of females/trap/10-days was 27.4 in August and 44.9 in
early September, after which it gradually decreased to 0.2 in early
December. The maximum number of females collected in a single
trap was 101. Moreover, few Culex pipiens ( n = 35) were caught.
The number of Ae. albopictus females collected daily in the
eight sticky traps is shown in Fig. 2 . The maximum and mini-
mum numbers of females collected were 25 and two, respec-
tively. During daily collections, we sometimes observed whole
sets of mosquito legs stuck to the adhesive sheets or areas with-
out glue in spots previously marked for the presence of mosqui-
toes. We concluded that mosquitoes had shed their legs when
trying to fly off from the trap, or, in the second instance, that the
bodies of mosquitoes had possibly been eaten by animals visit-
ing the traps.
Site 2 . By the end of the 2-year sampling period, the sticky
trap had collected 21 978 Ae. albopictus in 28 265 trap days.
Most of these (91.6%) were females, giving a mean collection
rate of 0.71 females/trap/day. The maximum mean numbers of
females/trap/week collected were 24.7 (week 5) and 20.2 (week
59) in the 2003 – 04 and 2004 – 05 seasons, respectively; between
weeks 19 and 41 and weeks 70 and 91 no mosquitoes were col-
lected. The maximum number of females collected in a single
trap was 83. A small number of Cx pipiens L. females ( n = 546)
and males ( n = 192), a few Culiseta longiareolata (Macquart),
Aedes geniculatus (Olivier), Aedes berlandi Séguy and a few
unidentified sandflies were also trapped.
Figure 3 shows the population dynamics of Ae. albopictus
adult females and eggs as obtained in sticky trap and ovitrap
catches, respectively. When traps were serviced every 2 – 4
weeks, the mean number of mosquitoes or eggs was divided by
the corresponding number of weeks separating two successive
sampling events. Overall, the sticky trap and the ovitrap yielded,
respectively, 0.6 Ae. albopictus females/trap/day and 6.1 eggs/
ovitrap/day in phase 1, and 1.0 females/trap/day and 9.0 eggs/
ovitrap/day in phase 2. Limiting the analysis to the peak seasons
(i.e. weeks 1–12 and 50–67 in phases 1 and 2, respectively,
which corresponded to periods when > 95% of the ovitraps
were found to have at least one egg), the sticky trap and the
ovitrap yielded 2.2 females/trap/day and 13.7 eggs/ovitrap/day,
respectively, in phase 1 and 1.9 females/trap/day and 19.4 eggs/
ovitrap/day, respectively, in phase 2. The results of the prelimi-
nary MRR experiment showed an overall recapture rate of
22.6% (i.e. 14 of the 62 Ae. albopictus females marked with
fluorescent pigments and released in the centre of site 2 were
recaptured within 2 weeks of being released, and 11 of them
were caught in the first week). No marked males were recap-
tured. Fluorescent individuals were found mainly in traps close
to the release site, but also in the most distant traps in the study
area, about 210 m from the release site.
Relationship between sticky trap and ovitrap catches
The logarithm for the mean number of Ae. albopictus females
caught during a week at site 2 by the sticky traps was highly
correlated with the logarithm for the mean number of eggs
found in the ovitraps (Pearson correlation coefficient r = 0.958,
95% confidence interval [CI] 0.932 – 0.974, n = 66, Student ‘ s
t = 15.3, P < 0.0001) ( Fig. 4 ). The correlation coefficient
remained homogeneous between the two sampling phases, at
r = 0.956 for the first period and r = 0.968 for the second
period ( ? 2 = 0.46, P = 0.79). On the logarithmic scale, the
slope of the fitted major axis regression line was 1.695 (95% CI
1.555 – 1.853) and the intercept 0.065 ( Fig. 4 ). The slope of the
major axis regression line on the logarithmic scale was signifi-
cantly different from 1 (its 95% CI did not overlap unity); hence
the functional relationship between the two sampling techniques
was density-dependent on the linear scale, with ovitraps collect-
ing proportionally more eggs than sticky traps collected mos-
quitoes at increasing population densities.
Fig. 2. Total numbers of Aedes albopictus
females collected daily with eight sticky traps
at site 1.
0
5
10
15
20
25
30
Total number of Aedes albopictus females
SeptemberOctober
681012
14
16
18
20 22 242628
30
246

Page 7

Development and evaluation of a sticky trap for container-breeding mosquitoes 189
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
Dispersion profile and minimum sample size
To assess the sampling properties of both the sticky trap and
the ovitrap and to provide a framework for optimal sampling
plans for Ae. albopictus , the dispersion profiles of sticky trap
and ovitrap catches were investigated by calculating the VMR,
and optimal sampling plans for each type of trap were quanti-
fied according to Taylor ’ s power law.
Dispersion profile . With the exception of four of 55 tests,
the VMR of weekly catches by the sticky traps was always sig-
nificantly > 1, indicating that most of the time the dispersion of
Ae. albopictus females was not compatible with a random distri-
bution. In the case of the ovitrap, all 64 tests returned a signifi-
cantly non-random distribution ( P < 0.05). After applying
Bonferroni correction for multiple tests, the numbers of non-
significant tests at the 5% experiment-wise error rate increased
to seven (12.7%) and four (6.3%) for sticky trap and ovitrap
catches, respectively. Weeks when the test returned a dispersion
profile compatible with a Poisson distribution covered mean
catches in the range of 0.05 – 1.35 mosquitoes/trap and 0.03 – 0.78
eggs/trap. Overall, the dispersion of Ae. albopictus adult fe-
males and eggs among the traps was highly aggregated, except
at the lowest range of densities measurable by our sampling
plan, when aggregation, even if present, was unlikely to be de-
tected. Thus, we assumed that the negative binomial distribution
constituted a good approximation of the dispersion of mosqui-
toes or eggs among the traps, as repeatedly found in numerous
studies of the spatial distribution of insects ( Kuno, 1991 ), and
proceeded to calculate the parameter k of this type of distribu-
tion. The calculation of a common k , however, was prevented by
a significant association between 1/ k and mean density in the
case of sticky trap catches ( F = 4.83, d.f. = 1,51, P = 0.03),
indicating that for this type of trap the level of aggregation
slightly decreases with mean density (slope of regression line re-
lating 1/ k and mean density, – 0.022 ± 0.010 standard error [SE]).
Hence, we estimated aggregation and optimal sampling plan pa-
rameters from the more general relationship between the vari-
ance and the mean of trap catches based on Taylor ’ s power law.
Taylor ’ s power law . The means and variance of weekly trap
catches were plotted and regression lines fitted to the datapoints
( Fig. 5 ). The regression coefficient was 1.325 (± 0.027 SE, 95%
CI 1.270 – 1.380) and 1.441 (± 0.034 SE, 95% CI 1.373 – 1.509)
for the sticky trap and ovitrap, respectively. As 95% CIs did not
overlap unity, both coefficients were significantly different from
1, at least at P < 0.05, indicating an aggregated distribution of
catches for both traps. The two coefficients were significantly
different from each other ( F = 40.51, d.f. = 1115, P < 0.0001),
indicating that the degree of aggregation was significantly
higher for ovitraps than for sticky traps. Back-transformation of
regression intercepts on the linear scale gave the following
Taylor ’ s power law equations: s 2 = 2.401 m 1.325 for the sticky
trap, and s 2 = 13.068 m 1.441 for the ovitrap.
Optimal sample plans . We used the estimates of the Taylor ’ s
power law coefficients to calculate: (a) the minimum number of
sample units necessary to obtain 95% CIs of mean density as ± d
of the sample mean (i.e., trap precision), and (b) the minimum

Fig. 4. Correlation between ovitrap and sticky trap mean weekly catches
in site 2. ◊ = sampling phase 1; ● = sampling phase 2 (cf. text). Both
variates on the abscissa and ordinate are subject to sampling error, which
is estimated by standard errors drawn on both axes for each point in the
scattergram. The fitted major axis regression is shown as a solid line.
Fig. 3. Population dynamics of Aedes albop-
ictus eggs and adult females in the zoological
gardens in Rome (site 2) during two succes-
sive breeding seasons (August 2003 – July
2005). Williams ’ means of sticky trap ( ◊ ) and
ovitrap ( ● ) catches are plotted starting from
week 1 (1 August 2003). Asterisks at data-
points denote weeks when the traps were
in place but were serviced on a 2 – 4-week
schedule; the density for these weeks was in-
terpolated as the mean value for the interval
separating two successive sampling events,
established from the mean catch of the first
week when the traps were serviced.

Page 8

190 L. Facchinelli et al.
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
number of sample units necessary to obtain at least one mos-
quito or egg on 95% of sampling occasions (hereafter trap sen-
sitivity ), according to formulae derived from optimal sampling
theory (see Materials and methods).
The minimum number of sample units necessary to maintain
a fixed level of precision or sensitivity depends upon the mean
density of the population to be sampled: at lower population
densities, the sampling effort required to keep a certain degree
of precision or sensitivity increases monotonically. The func-
tional relationship between mean density and number of sample
units is plotted for three levels of precision in Fig. 6a for the
sticky trap and Fig. 6b for the ovitrap. As expected, we found
that high levels of precision required a large sampling effort at
low population densities. For example, Fig. 6a shows that in the
case of the sticky trap, 1092 sampling units at a mean density of
0.1 are required to obtain estimates of the sample mean within
± 20% of the parametric mean with 95% probability. Similarly,
Fig. 6b shows that in the case of the ovitrap, 1255 sampling
units at a density of one egg per ovitrap are needed to achieve
the same level of precision. These are often impractical levels of
sampling effort and more feasible targets of accuracy can be
arbitrarily established by setting ad hoc values for the parame-
ters ? and d in equation 1 . However, the minimum number of
sample units becomes feasible at the range of population densi-
ties that are currently observed in medium- to highly infested
Fig. 6. Trap precision. Relationship between
the mean density in the population, expressed
as the weekly mean number of Aedes albopic-
tus adult females or eggs per trap, and the
minimum number of sample units necessary
to obtain 95% confidence limits as a fixed pro-
portion ± d of the sample mean. Three curves
are drawn for values of d = 10%, 20% or 30%
for both the sticky trap (a) and ovitrap (b).

Fig. 5. Fitted Taylor ’ s power law regression lines for Aedes albopictus
mean weekly catches by the sticky trap ( ◊ ) or standard ovitrap ( ● ).

Page 9

Development and evaluation of a sticky trap for container-breeding mosquitoes 191
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
areas during periods of stable dynamics. At site 2, we frequently
recorded catches of c. 10 Ae. albopictus females per sticky trap
and c. 100 eggs per ovitrap ( Figs 3 and 5 ). At these population
densities, the minimum number of sample units required for a
30% level of precision are n = 22, and n = 42 units, respec-
tively, representing quite reasonable figures.
Figure 7 shows the relationship between the minimum number
of sample units and the mean density for a sampling plan requir-
ing a 95% level of sensitivity. Again, the sampling effort needed
is prohibitively large at very low densities ( ≤ 0.001). At a popu-
lation density of 0.01, our estimates give n = 223 sample units
for the sticky trap, and n = 397 sample units for the ovitrap.
These figures drop to 32 and 72, respectively, at a mean density
of 0.1.
However, the units of density are inherently different between
the two traps (i.e. one adult female for the sticky trap and one
egg for the ovitrap). Thus, in order to compare the sampling
properties of the two traps, we calculated calibration curves for
the ovitrap, the units of which were standardized according to
the functional relationship between the two sampling devices,
as assessed by major axis regression ( Fig. 4 ). The resulting cali-
bration curves are plotted in Fig. 8 . Figure 8a shows the calibra-
tion curve calculated at d = 30%; the plot demonstrates that, at
exceedingly low ( ? < 0.02) or high ( ? > 400) mosquito densi-
ties, the ovitrap is comparatively more precise than the sticky
trap. Within the range of densities encountered in our study
area, however, the sticky trap required a lower number of sam-
ple units than the ovitrap for the same degree of precision. The
same pattern applies for other degrees of precision, d .
At population densities corresponding to a threshold of
? 0.135 females per sticky trap (i.e. ? 0.440 eggs per ovitrap),
the two sampling techniques require the same number of traps
( n = 25) to achieve a 95% level of sensitivity ( Fig. 8b ). At densities
higher than this, the sticky trap needs fewer sampling units than
the ovitrap to achieve the same degree of sensitivity (i.e. if every-
thing else is equal, the sticky trap is a more sensitive sampling
tool than the ovitrap at this range of densities). However, at den-
sities increasingly lower than this threshold, the ovitrap requires
a markedly lower number of sampling units.
Discussion
Sampling properties of the sticky trap
Under our ecological conditions, the sticky trap we devised
was effective and highly specific for collecting Ae. albopictus ,
which represented 96.8% of the total mosquitoes collected in
either site, despite the presence of other mosquito species in
Rome ( Romi & Sabatinelli, 1997 ). In fact, only a few Cx pipiens ,
Cu. longiareolata , Ae. berlandi and Ae. geniculatus specimens
were captured in the trap. In particular, it is relevant to note
that at site 2, Cx pipiens represented 3.2% of the mosquitoes
collected in the sticky trap, despite the findings of Pombi et al.
(2003) , showing that the larval densities of this species in the
sewer catch basins of the area were markedly higher (i.e. 0 – 33.2
larvae/dip) than those of Ae. albopictus (i.e. 0 – 1.3 larvae/dip).
It was relatively easy to assess the taxon and gender of cap-
tured specimens directly by eye in the field, even without the
help of a magnifying glass. However, we anticipate that in areas
where Stegomyia mosquitoes of similar morphology are present
(e.g. Ae. aegypti , Ae. albopictus , Ae. scutellaris ), species identi-
fication under a dissecting microscope is advisable. In fact, we
already have records showing that, in different geographical ar-
eas (i.e. central Thailand, northern Queensland, Burkina Faso
and Venezuela), the sticky trap captures container-breeding
mosquitoes other than Ae. albopictus , such as Ae. aegypti, the
major vector of yellow fever and dengue, and Culex quinquefas-
ciatus , a vector of other arboviruses and filariae (Facchinelli
et al. , 2005, unpublished data; Valerio et al. , 2005, unpublished
data; Costantini et al. , 2006 unpublished data).
The number of mosquitoes on the sticky sheets was easily
counted directly in the field; however, it is possible that the
catch was sometimes underestimated as a result of small geckos,
lizards and snails having eaten the bodies of captured mosqui-
toes, leaving sets of legs on the glue, as shown by the recurrent
observation of whole sets of mosquito legs without bodies stuck
on the adhesive sheets.
As expected, the majority of females collected were gravid,
although we frequently also noticed the presence of unfed and,
Fig. 7. Trap sensitivity. Relationship be-
tween the mean density in the population,
expressed as the weekly mean number of Ae-
des albopictus adult females or eggs per trap,
and the minimum number of sample units
necessary to obtain at least one adult female
Ae. albopictus with the sticky trap (solid
line), or Ae. albopictus egg with the ovitrap
(dotted line), on 95% of sampling occasions.

Page 10

192 L. Facchinelli et al.
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
more rarely, blood-fed females, suggesting that the traps were
exploited not only by ovipositing females, but also by other
fractions of the mosquito population, probably in response to
cues associated with resting sites. This hypothesis is supported
by the repeated presence of males in the traps, although they
represented < 10% of the total number of Ae. albopictus
collected.
Both the 10-day and weekly schedules of trap servicing were
adequate under our conditions. The lid on the top of the trap
protected the mosquitoes on the sticky sheets from rain and sun-
shine, thereby preserving both captured specimens and glue in
good condition. However, the optimal frequency of trap servic-
ing might depend on local ecological and environmental condi-
tions. We anticipate that, under certain conditions, debris might
accumulate on the sticky sheets, hindering the identification of
captured mosquitoes and the adhesive properties of the glue.
Moreover, as we occasionally found some larvae breeding in the
trap, a weekly servicing would be advisable under climatic con-
ditions favourable for rapid larval development. We also hy-
pothesize that rains stronger than those experienced during our
trial might wet and damage the captured specimens. A larger lid
could be provided in these cases to protect the adhesive surfaces
from heavy rains.
Comparison of sticky trap and ovitrap catches
At site 2, the mean number of Ae. albopictus females col-
lected by the sticky traps was highly correlated with the mean
number of eggs collected by the ovitraps over a wide range
of densities. Despite the operational differences between the
two sampling phases (the sampling plan changed from a sticky
trap : ovitrap ratio of 3 : 1 to one of 1 : 1 between phases; see
Materials and methods), the degree of correlation between the
two sampling tools remained homogenously high. A linear rela-
tionship between the mean number of eggs collected by the
ovitrap and the mean number of females collected by the sticky
trap was fitted on the logarithmic scale ( Fig. 4 ). This model al-
lows for the description of the density-dependent relation-
ship between the mean number of females/sticky trap and the
Fig. 8. Standardized calibration curves for
95% trap sample precision (a) and sensitivity
(b). In (a), the curves are calculated for d =
30%. Solid lines refer to the sticky trap, and
are identical to those plotted in Figs 6a and 7
for this sampling device. Dotted lines refer to
the calibration curves for the ovitrap.

Page 11

Development and evaluation of a sticky trap for container-breeding mosquitoes 193
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
mean number of eggs/ovitrap under our experimental conditions.
From the log-regression model, it can be estimated that when
the sticky trap collects one or 10 Ae. albopictus females/trap,
the ovitrap is expected to catch 2.8 or 66.6 eggs/ovitrap, respec-
tively. Thus, no single multiplication factor on the linear scale
can relate these two variates.
Both types of trap yielded highly heterogeneous catches, as
demonstrated by the high VMRs, indicating that some traps
caught greater numbers of mosquitoes or eggs, whereas others
were markedly less productive. Taylor ’ s power law was fitted to
the trap samples, providing evidence that the degree of aggrega-
tion was significantly higher in the case of eggs per ovitrap
compared with mosquitoes per sticky trap. Given the aggregated
pattern of the trap samples, we used the coefficients of Taylor ’ s
power law regression lines ( Fig. 5 ) to develop optimal sampling
plans (equations 1 − 3). The usefulness of this conceptual frame-
work is that optimal sampling plans can set a priori levels of
statistical precision required from sample estimates (e.g. when
levels of infestation must be compared among locations, or be-
fore and after a vector control intervention). For example, sup-
pose that a vector control operation is planned in an area where
Ae. albopictus densities are expected to average 10 adult fe-
males per sticky trap. Vector control managers may wish to
verify that they can obtain a reduction of at least 50% in mos-
quito densities, and want to be able to detect whether their con-
trol intervention is successful in attaining this target or otherwise.
Thus, the sampling plan required to evaluate the impact of the
control operation must be set so that there is sufficiently high
statistical power (say, 95% probability) to be able to detect a
significant difference between the mean densities before and
after treatment. One way to do so is to calculate (using equation 1 )
the number of sampling units that will return estimates of mean
density before and after treatment whose 95% CIs do not over-
lap. With a fixed precision of 30%, the sample estimate of
mean density will lie in the range of 7 – 13 before treatment, and
3.5 – 6.5 post-treatment if the control operation attains its target.
By putting the relevant parameters in equation 1 (i.e. d = 0.3,
? = 0.95, m = 10 before the control intervention, and m = 5
after the control intervention), one can calculate a priori the min-
imum number of sample units required to attain the level of sta-
tistical precision needed to test this hypothesis. In this fictitious
case, n should be 22 before and 35 after the control operation.
Under certain operational conditions it is more appropriate to
estimate the minimum number of samples required to give a
positive record with a known level of probability. This is often
the case in surveillance campaigns where only the presence of a
potential invader species needs to be ascertained, not its density.
In these conditions, the invader will often be present at very low
densities during the early stages of colonization. Thus, good
sensitivity of the sampling plan, rather than precision, is re-
quired in this context.
Our results show that, at population densities higher than
0.135 females/trap, our novel sticky trap is more sensitive than
the ovitrap in detecting the presence of Ae. albopictus ( Fig. 8b ),
(i.e. a lower number of sticky traps than ovitraps is needed to
catch at least one mosquito [vs. one egg] on 95% of sampling
occasions). At densities below this threshold, the ovitrap is more
sensitive than the sticky trap. This reversal is presumably the
outcome of the different sampling properties and unknown rela-
tive efficacy of the two trapping devices, whereby the density-
dependent functional relationship relating the two sampling
devices and differences in dispersion profiles contribute to shape
the observed pattern.
Conclusions
Overall, our results indicate that our new model of sticky trap
can be used effectively to catch large numbers of Ae. albopictus
adults, and as an alternative sampling tool to the standard ovit-
rap for monitoring the population dynamics of Ae. albopictus
and, possibly, other container-breeding mosquitoes, particularly
those belonging to the subgenus Stegomyia ( e.g. Ae. aegypti ).
Moreover, our results indicate that the sticky trap is more sensi-
tive than the ovitrap for detecting the presence of Ae. albopictus
during its reproductive season in Rome, although the opposite
was shown at very low densities. It will therefore be important
to determine the sensitivity of the sticky trap before using it in
surveillance activities under other ecological situations.
The sticky trap has several operational advantages over the
ovitrap: (a) in areas of sympatry of Aedes whose eggs are mor-
phologically indistinguishable, specimens captured with the
sticky trap can be easily identified, whereas the ovitrap requires
that eggs are kept for the identification of larvae or adults, which
represents a major constraint in quarantine surveillance opera-
tions, where the immediate identification of introduced pests is
desirable; (b) specimens collected by the sticky trap can be
counted relatively easily, eliminating the need for egg-counting
under a microscope, and (c) the sticky trap overcomes the inher-
ent limitation of trapping devices that collect eggs rather than
adult mosquitoes: the latter do not allow a simple extrapolation
of the number of ovipositing females from the number of eggs
collected. However, the efficiency of the sticky trap is likely to
be affected by the availability and abundance of alternative
breeding sites in the surveyed area, as has already been demon-
strated with the ovitrap ( Focks, 2003 ).
Compared with other types of sampling device targeted at
trapping host-seeking adult mosquitoes (e.g. the Fay − Prince,
CDC Wilton and BG-Sentinel traps), the sticky trap is cheap,
easy to manipulate and does not need a source of power.
However, it should be stressed that as the sticky trap collects
mainly gravid females, it could be employed for the surveil-
lance of container-breeding mosquito pests and vectors, and/or
the monitoring of the impact of mosquito control tactics, but not
for the estimation of man − vector contact. Compared with back-
pack aspirators which are used periodically (e.g. for a few min-
utes/week in each sampling premise) to actively collect
mosquitoes, the sticky trap operates for the whole sampling
period and is not biased by the operator ’ s skill level.
The sticky trap also represents an attractive tool for ecologi-
cal and epidemiological research. The specimens collected by
the sticky trap can be used for studies on dispersal and longev-
ity: the results of our preliminary MRR experiment confirm that
specimens dusted with fluorescent powders can be detected
relatively easily by observing the sticky sheets under UV light,
as has already been shown with a different type of sticky trap

Page 12

194 L. Facchinelli et al.
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
( Russell et al. , 2005 ). It is interesting to note that, whereas we
obtained a recapture rate of 22.6%, Russell et al. (2005)
reported a recapture rate after 15 days of only 3.4% for Ae.
aegypti , although their trial was carried out in a study area
of comparable size and with a much higher density of traps (9.2
traps/ha vs. 3.8 traps/ha. Differences in ecological conditions
and target species may account for these results. Although we
did not carry out such specific tests in our study, we anticipate
that the specimens collected by our sticky trap could also be: (a)
assayed for the presence of pathogens, such as the dengue virus
( Bangs et al. , 2001; Ritchie et al. , 2004 ); (b) genotyped for in-
secticide resistance alleles ( Rodriguez et al. , 2001 ), and (c) ana-
lysed for the origin of the bloodmeal in the case of specimens
captured before the completion of bloodmeal digestion. Finally,
we suggest that, in areas where alternative breeding habitats or
resting sites are scarce, our sticky trap may potentially be used
as a mass trapping device for vector control and integrated pest
management.
Further studies are in progress to evaluate the sampling ca-
pacity of the sticky trap in Ae. aegypti -infested areas (where this
species represents the main dengue vector), in either the pres-
ence or absence of oviposition attractants, which might increase
the sticky trap catch ( Reiter et al. , 1991; Ritchie, 2001 ).
Acknowledgements
Financial support was provided by SANAMA s.r.l. We thank
Bioparco s.p.a. (Rome, Italy) for hospitality, Antonella Daidone,
Fulvio Fraticelli, Walter Iorio, Luciano Lentini and Fulvio
Torreti for their support and encouragement during field work;
Caterina Fanelo, Vincenza Petrarca and Scott Richie for reading
and commentating on the manuscript, and Mario Coluzzi for
stimulating the development and testing of the trap.
References
Bangs , M.J. , Tan , R. , Listiyaningsih , E. , Kay , B.H. & Porter , K.R . ( 2001 )
Detection of dengue viral RNA in Aedes aegypti (Diptera: Culicidae)
exposed to sticky lures using reverse-transcriptase polymerase chain
reaction . Journal of Medical Entomology , 38 , 720 – 724 .
Buntin , G.D . ( 1994 ) Developing a primary sampling program . Hand-
book of Sampling Methods for Arthropods in Agriculture ( ed . by L.P.
Pedigo & G.D. Buntin ), pp . 99 – 115 . CRC Press , Boca Raton .
Chen , C.D. , Benjamin , S. , Saranum , M.M. , Chiang , Y.F. , Lee , H.L. ,
Nazni , W.A. & Sofian Azirun , M . ( 2005 ) Dengue vector surveillance
in urban residential and settlement areas in Selangor, Malaysia . Trop-
ical Biomedicine , 22 , 39 – 43 .
Clark , G.G. , Seda , H. & Gubler , D.J . ( 1994 ) Use of the ‘ CDC backpack
aspirator ’ for surveillance of Aedes aegypti in San Juan, Puerto Rico .
Journal of the American Mosquito Control Association , 10 , 119 – 124 .
Clements , A.N . ( 1999 ) The Biology of Mosquitoes , Vol. 2 : Egg Laying ,
pp . 623 – 625 . CABI Publishing , Wallingford, U.K .
Di Luca , M. , Toma , L. , Severini , F. , D ’ Ancona , F. & Romi , R . ( 2001 )
[ Aedes albopictus in Rome: monitoring in the 3-year period of 1998 –
2000 .] [ In Italian .] Annali dell ’ Istituto Superiore di Sanita , 37 , 249 – 254 .
Elliott , J.M . ( 1977 ) Some methods for the statistical analysis of samples
of benthic invertebrates . Scientific Publications of the Freshwater
Biological Association , 25 , 1 – 156 .
Fay , R.W. & Eliason , D.A . ( 1966 ) A preferred oviposition site as a sur-
veillance method for Aedes aegypti . Mosquito News , 26 , 531 – 535 .
Focks , D.A . ( 2003 ) A Review of Entomological Sampling Methods and
Indicators for Dengue Vectors . [ WHO document: TDR/IDE/Den/
03 1.] World Health Organization , Geneva .
Gratz , N.G . ( 2004 ) Critical review of the vector status of Aedes albopic-
tus . Medical and Veterinary Entomology , 18 , 215 – 227 .
Hawley , W.A . ( 1988 ) The biology of Aedes albopictus . [Review.] Jour-
nal of the American Mosquito Control Association , 1 ( Suppl ), 1 – 39 .
Kuno , E . ( 1991 ) Sampling and analysis of insect populations . Annual
Review of Entomology , 36 , 285 – 304 .
Liew , C. & Curtis , C.F . ( 2005 ) Horizontal and vertical dispersal of
dengue vector mosquitoes, Aedes aegypti and Aedes albopictus , in
Singapore . Medical Veterinary Entomology , 18 , 351 – 360 .
Mercado Hernandez , R. , Aguilar Gueta , J.D. , Fernandez Salas , I. &
Earl , P.R . ( 2006 ) The association of Aedes aegypti and Ae. albopictus
in Allende, Nuevo Leon, Mexico . Journal of the American Mosquito
Control Association , 22 , 5 – 9 .
Ordonez Gonzalez , J.G. , Mercado Hernandez , R. , Flores Suarez , A.E.
& Fernandez Salas , I . ( 2001 ) The use of sticky ovitraps to estimate
dispersal of Aedes aegypti in northeastern Mexico . Journal of the
American Mosquito Control Association , 17 , 93 – 97 .
Pitcairn , M.J. , Wilson , L.T. , Washino , R.K. & Rejmankova , E . ( 1994 )
Spatial patterns of Anopheles freeborni and Culex tarsalis (Diptera:
Culicidae) larvae in Californian rice fields . Journal of Medical Ento-
mology , 31 , 545 – 553 .
Pombi , M. , Costantini , C. & Della Torre , A . ( 2003 ) Aedes albopictus
(Diptera: Culicidae) in Rome: experimental study of relevant control
strategy parameters . Parassitologia , 45 , 97 – 102 .
Reiter , P. , Amador , M.A. & Nelson , C . ( 1991 ) Enhancement of the CDC
ovitrap with hay infusions for daily monitoring of Aedes aegypti
populations . Journal of the American Mosquito Control Association ,
7 , 52 – 55 .
Reiter , P. & Nathan , M.B . ( 2001 ) Guidelines for Assessing the Efficacy
of Insecticidal Space Sprays for Control of the Dengue Vector Aedes
Aegypti . [ WHO document: WHO/CDS/CPE/PVC/2001.1 .] World
Health Organization , Geneva .
Ritchie , S.A . ( 2001 ) Effects of some animal feeds and oviposition sub-
strates on Aedes oviposition in ovitraps in Cairns, Australia . Journal
of the American Mosquito Control Association , 17 , 206 – 208 .
Ritchie , S.A. , Long , S. , Hart , A. , Webb , C.E. & Russell , R.C . ( 2003 ) An
adulticidal sticky ovitrap for sampling container-breeding mosquitoes .
Journal of the American Mosquito Control Association , 19 , 235 – 242 .
Ritchie , S.A. , Long , S. , Smith , G. , Pyke , A. & Knox , T.B . ( 2004 ) Ento-
mological investigations in a focus of dengue transmission in Cairns,
Queensland, Australia, by using sticky ovitraps . Journal of Medical
Entomology , 41 , 1 – 4 .
Ritchie , S.A. , Moore , P. , Carruthers , M . et al . ( 2006 ) Discovery of a wide-
spread infestation of Aedes albopictus in the Torres Strait, Australia .
Journal of the American Mosquito Control Association , 22 , 358 – 365 .
Rodriguez , M.M. , Bisset , J. , de Fernandez , D.M. , Lauzan , L. & Soca , A .
( 2001 ) Detection of insecticide resistance in Aedes aegypti (Diptera:
Culicidae) from Cuba and Venezuela . Journal of Medical Entomol-
ogy , 38 , 623 – 628 .
Romi , R. & Sabatinelli , G . ( 1997 ) Diptera: Culicidae. Insetti di Roma
( ed . by M. Zapparoli ), pp . 246 – 248 . Fratelli Palombi Editori , Rome .
Rozeboom , L.E. , Rosen , L. & Ikeda , J . ( 1973 ) Observations on oviposi-
tion by Aedes (S.) albopictus Skuse and Ae. (S.) polynesiensis Marks
in nature . Journal of Medical Entomology , 10 , 397 – 399 .

Page 13

Development and evaluation of a sticky trap for container-breeding mosquitoes 195
© 2007 The Authors
Journal compilation © 2007 The Royal Entomological Society, Medical and Veterinary Entomology, 21, 183–195
Russell , R.C. & Ritchie , S.A . ( 2004 ) Surveillance and behavioural
investigation of Aedes aegypti and Aedes polynesiensis in Moorea,
French Polynesia, using a sticky ovitrap . Journal of the American
Mosquito Control Association , 20 , 370 – 375 .
Russell , R.C. , Webb , C.E. , Williams , C.R. & Ritchie , S.A . ( 2005 )
Mark − release − recapture study to measure dispersal of the mosquito
Aedes aegypti in Cairns, Queensland, Australia . Medical and Veteri-
nary Entomology , 19 , 451 – 457 .
Scott , T.W. & Morrison , A . ( 2003 ) Aedes aegypti density and the risk
of dengue virus transmission . Ecological Aspects for Application of
Genetically Modified Mosquitoes ( ed . by W. Takken & T.W. Scott ),
pp . 187 – 206 . Kluwer Academic Publishers , Dordrecht .
Service , M.W. ( 1993 ) Mosquito Ecology Field Sampling Methods , 2nd
edn , pp . 515 – 520 . Elsevier Applied Science , London .
Sokal , R.R. & Rohlf , J.F . ( 1995 ) Biometry , 3rd edn , pp . 686 – 697 . W.H.
Freeman , New York .
Taylor , L.R . ( 1961 ) Aggregation, variance and the mean . Nature , 189 ,
732 – 735 .
Taylor , L.R . ( 1984 ) Assessing and interpreting the spatial distributions
of insect populations . Annual Review of Entomology , 29 , 321 – 357 .
Vezzani , D. , Velazquez , S.M. & Schweigmann , N . ( 2004 ) Control of
Aedes aegypti with temephos in a Buenos Aires cemetery, Argentina .
Revista de Saude Publica , 38 , 738 – 740 .
Accepted 18 March 2007