Highlights

Household scale Wolbachia release strategies for effective dengue

control
Abby Barlow, Ben Adams
arXiv:2507.23636v1 [q-bio.PE] 31 Jul 2025

• We developed a simulation model for the movement of mosquitoes be-

tween households.

• We investigated the release of Wolbachia-infected mosquitoes at both

the household and community scale.

• Household scale releases result in rapid and successful invasion into the
household and the wider community.

• When regular releases are carried out in a single household it can pro-
vide protection for the entire community of households if releases are
carried out frequently for a number of years and a sufficient number of
females are released on each release event.
 Household scale Wolbachia release strategies for
effective dengue control
Abby Barlowa , Ben Adamsa
a
Department of Mathematical Sciences, The University of Bath, Bath, BA2
7AY, Somerset, United Kingdom

Abstract
The release of Wolbachia-infected mosquitoes into Aedes aegypti infested ar-
eas is a promising strategy for localised eradication of dengue infection. Ae
aegypti mosquitoes favour urban environments as breeding habitats, so are
often found in and around houses. Therefore, it is likely that they will in-
fect members of the households that they reside around. Since population
groupings within households are small, stochastic effects become important.
Despite this, little work has been carried out to investigate the outcome of
releasing Wolbachia-infected mosquitoes at a household scale, either from an
empirical and theoretical stand point. In previous work, we developed and
analysed a stochastic (continuous time Markov chain) model for the invasion
of Wolbachia-infected mosquitoes into a single household containing a pop-
ulation of wildtype mosquitoes. In the present study, we extend our frame-
work to a connected community of households coupled by the movement
of mosquitoes. We use numerical results obtained via Gillespie’s stochas-
tic simulation algorithm to investigate optimal strategies for the release of
Wolbachia-infected mosquitoes carried out at either the community or the
household scale. We find that household scale releases can facilitate rapid
and successful invasion of the Wolbachia-infected mosquitoes into the house-
hold population and then into the wider community. We further explore the
impact of regular household scale releases of Wolbachia-infected mosquitoes
for a range of compositions for the release population, time intervals between
releases and proportion of households participating in the releases. We find
that a single release household can provide sufficient protection to the entire
community of households if releases are carried out frequently for a number
of years and a sufficient number of females are released on each occasion.
Keywords: Wolbachia, household model, stochastic, invasion, release
strategies, movement, metapopulation, dengue

1. Introduction
Dengue is a common vector-borne disease which is transmitted to humans
via the bite of an infected mosquito WHO. It is a neglected tropical disease
that is present in over 100 countries and accounts for around 40, 000 deaths
worldwide each year (Hasan et al., 2016; Bhatt et al., 2013; Zeng et al., 2021).
The case burden is estimated to range between 100 and 400 million per year
(Zeng et al., 2021). The development of vaccinations against dengue has been
challenging. Dengvaxia is currently the only fully licensed vaccination and
is only recommended for individuals who have already been infected with
at least one serotype (WHO). Therefore, efforts to prevent dengue infection
often target the vector populations. The most common carrier, the Aedes
aegypti mosquito, inhabits urban and suburban areas, breeding in artificial
water containers. These breeding sites are often found in and around homes.
Insecticides are sprayed both indoors and outdoors to control the popula-
tion. However, the method is unsustainable in the long-term since continual
spraying is often needed to completely exterminate the mosquito population.
This can result in the mosquitoes building up resistance to insecticides.
Another well established approach to mosquito control is to introduce a
large number of sterilised male mosquitoes into the local population. Often
this is implemented by passing small levels of ionizing radiation over the pop-
ulation to be introduced (Strugarek et al., 2019). A more recent sterile insect
approach is based on the release of genetically modified male mosquitoes who
carry a lethal dominant gene; this technique is known as RIDL (Dorigatti
et al., 2018). An alternative control method utilises the intracellular bacte-
ria, Wolbachia. Wolbachia does not cause any harm to humans and is found
naturally in around 60% of insects (CDC, 2022). Infection alters the biology
of Ae aegypti in numerous ways that can reduce the prevalence of infection.
Firstly, with certain Wolbachia strains developed for Ae aegypti transinfec-
tion, a Wolbachia-infected male and a wildtype female cannot produce viable
offspring. In contrast, the offspring of an infected male and an infected fe-
male are viable and, with a certain probability, Wolbachia positive. This
mechanism is called cytoplasmic incompatibility (CI) (Dorigatti et al., 2018)
and affords infected females a reproductive advantage over wildtypes. Sec-
ondly, Wolbachia infection can impose fitness costs on the mosquitoes, such

2
as decreased lifespan and fecundity and prolonged larval development. Fi-
nally, Wolbachia infection has been shown to reduce the ability of mosquitoes
to transmit infection via their salivary glands (Hughes and Britton, 2013).
Many empirical studies have investigated how Wolbachia-infected mosquitoes
invade wildtype populations and can be used as a control measure against
dengue (Ross, 2021; Dos Santos et al., 2022; O’Neill et al., 2018; Hoffmann
et al., 2014; Dufault et al., 2022). Cage experiments have shown that certain
strains are able to invade wildtype populations and establish in large numbers
(McMeniman et al., 2009; Walker et al., 2011). Numerous field releases have
also been implemented. In a small community near Cairns, Australia, in 2012
almost 300, 000 mosquitoes infected with the wMel Wolbachia strain were re-
leased over 10 weeks. The Wolbachia-infected mosquito frequency peaked at
90% in the local population (Hoffmann et al., 2011). Field releases of the
wMelPop strain (distinct from the wMel strain) were implemented in Tri
Nguyen, Vietnam (2013) (Nguyen et al., 2015). In this study, they released
pupae instead of adult mosquitoes over a period of 23 months. The Wol-
bachia frequency exceeded 90% by the end of the release period. However,
approximately 40 to 60 weeks later the frequency dropped to 20%, suggesting
that the strain’s high fitness cost renders it unsustainable in the absence of
regular re-introduction. More recently, large scale releases have been carried
out in Indonesia, Vietnam, Australia, Columbia and Brazil. Unlike previous
studies, these experiments focussed on determining the viability of the wMel
strain as a dengue control measure (Dorigatti et al., 2018; Dufault et al.,
2022; O’Neill et al., 2018; Dos Santos et al., 2022). They found that releas-
ing mosquitoes infected with this Wolbachia strain successfully interrupted
focal dengue virus transmission and reduced the incidence of human infection
in the area.
Mathematical modelling has played an important role in the development
of Wolbachia control by characterising the eco-epidemiological dynamics and
untangling the relationships observed in data. Many previous models have
used ODEs to describe the changes in density of the wildtype and Wolbachia-
infected mosquito populations (Hughes and Britton, 2013; Ndii et al., 2015).
These models often assume homogeneous, well-mixed populations and may
account for additional aspects such as multiple mosquito life stages and sea-
sonal fluctuations (Ndii et al., 2015). These relatively simple models incor-
porate human infection dynamics and use the infection basic reproduction
number R0 as an indicator for dengue control. The authors of Hughes and
Britton (2013) found that when Wolbachia infection is able to invade and

3
persist in the population, it has the potential to eliminate dengue when R0
is low. However, (Ndii et al., 2015) analysed a model parameterised for the
wMel strain and found that Wolbachia infection did not reduce mosquito
lifespan sufficiently to affect population persistence.
The authors of Dye and Cain (2024) adapted an ODE framework to pro-
duce a metapopulation model that accounts for spatially discrete mosquito
habitats coupled via mosquito movement. The population dynamics in each
habitat are described by a system of ODEs and movement between habitats
is modelled using dispersal kernels. They deduce that under certain condi-
tions, the introduction of Wolbachia-infected mosquitoes into a single habitat
can elicit persistent establishment across all the habitats. Other models use
reaction diffusion PDEs and dispersal kernels to model the movement of
mosquitoes across continuous space (Turelli and Barton, 2017). Such mod-
els focus on whether the Wolbachia-infected mosquitoes are able to diffuse
over a target area. In Turelli and Barton (2017) they found that, with the
wMel strain, even very slow spatial spread can lead to area-wide population
invasion within a few years of release.
Simulation based models permit a more finely detailed modelling frame-
work. The authors of Hancock et al. (2019) developed a metapopulation
model of mosquito habitats with heterogeneous density dependent competi-
tion effects and juvenile mosquito life stages. They found that varying the
‘habitat quality’ spatially can significantly impact Wolbachia invasion across
space. The authors of Pagendam et al. (2020) simulated a Markov process
model accounting for multiple mosquito life stages and multiple releases.
They found that under a full mosquito elimination strategy, dynamically
adapting male Wolbachia-release sizes can reduce cost and the risk of releas-
ing females.
In a previous study Barlow et al. (2025a) we introduced a stochastic
mathematical model for the dynamics of Wolbachia invasion in a single house-
hold (or a population of unconnected households) using a continuous time
Markov chain framework. We examined the relationship between the inva-
sion probability and the Wolbachia-infection population proportion under
the stochastic dynamics associated with small households, and compared
this with corresponding deterministic invasion thresholds. Here we extend
the model to consider a community of households via stochastic simulation
methods. Each household contains (within it or close by) a breeding habitat
for Ae aegypti mosquitoes. We assume mosquitoes bite the human members
of their household, so that the invasion dynamics unfold stochastically in the

4
metapopulation composed of a large number of small populations connected
by the movement of mosquitoes. This framework allows for localised inva-
sion, reversion and reinvasion, heterogeneously across the community. We
use this model to investigate the factors that determine invasion success,
and the impact of different strategies for the release of Wolbachia-infected
mosquitoes.
When designing a Wolbachia release strategy, there are numerous aspects
to consider. One key factor is the scale of the release. Previously, most em-
pirical studies have focused on performing releases at a community wide scale
(Dos Santos et al., 2022; Hoffmann et al., 2014; Dufault et al., 2022). Here,
one or more large populations are released in public spaces and then disperse
into residential areas. An alternative is to perform releases at a household
scale. Here each household is responsible for performing a small release on
their own property. We have found some evidence of this strategy being
carried out in practice (O’Neill et al., 2018). We hypothesize that house-
hold scale releases will lead to a faster invasion of the Wolbachia-infected
mosquitoes. Another key factor is the composition of the release population.
Releasing female Wolbachia-infected mosquitoes can result in replacement
of the wildtype population by a Wolbachia-infected population if the fitness
costs of Wolbachia infection are not too high. This wildtype replacement
outcome may be desirable since Wolbachia-infection reduces the transmis-
sion competence of mosquitoes for dengue and some other viruses. However,
releasing female mosquitoes may be undesirable if it temporarily increases
the vectorial capacity. In contrast, releasing only male Wolbachia-infected
mosquitoes can result in suppression of the total mosquito population size,
possibly to the point of eradication (Ross, 2021). This outcome may ap-
pear more desirable than replacement but mosquito-free households are at
immediate risk of reinvasion from regional wildtype populations. A third
possibility is a strategy in which both male and female infected mosquitoes
are released in ratios that vary over time.
Here, we consider a population of households connected by the movement
of mosquitoes. Female Ae aegypti mosquitoes typically move less than 100m
in their lifetime. Therefore it may be appropriate to assume they spend all
or most of their time within the same household. In contrast, males have
been observed to rapidly disperse up to 400m from their initial release site
(Trewin et al., 2021), though physical barriers such as residential blocks are
thought to impede their movement (Trewin et al., 2021) and field studies have
estimated the mean distance travelled over their lifetimes to range between

5
196m and 294m (Juarez et al., 2020; Marcantonio et al., 2019; Trewin et al.,
2020). These contrasting behaviours are likely in part because females seek
regular blood meals in addition to places to rest for oviposition, whereas
males spend most of their time searching for a mate.
We use a stochastic simulation framework (Gillespie, 1977) to explore the
efficacy of a range of single and regular release scenarios. We focus on several
key aspects of release design: (1) the scale the release is performed at; (2)
the number and sex ratio of the released Wolbachia-infected mosquitoes; (3)
the time between releases and (4) the proportion of households participating
in releases. In Section 2, we define our model framework, and the release
scenarios we carry out. In Section 3, we present our simulation results.
In Subsection 3.1 we study the efficacy of performing a single community
wide release in contrast to highly localised releases in each household. In
Subsection 3.2 we consider household and community scale releases composed
only of male Wolbachia-infected mosquitoes. In Subsection 3.3, we analyse
the impact of the number and ratio of male and female Wolbachia-infected
mosquitoes released into half of the households in the community in a single
release event. In Subsection 3.4 we investigate the outcome of regularly
releasing Wolbachia-infected mosquitoes into a single household.
We believe that this study will be valuable in informing effective and
efficient localised Wolbachia release strategies as a control against dengue
infection and other vector-borne diseases.

2. The model
In this section, we describe our model framework. Let the number of
male wildtype and Wolbachia-infected mosquitoes in a household be denoted
by mM and wM respectively. Analogously, the numbers of female wildtype
and Wolbachia-infected mosquitoes in a household are denoted mF and wF .
The state of each household is denoted by the resident mosquito population
array (mM , mF , wM , wF ). A household can transition to a new state via a
birth, death or movement event. The rates at which the population of male
wildtype and Wolbachia-infected mosquitoes die are respectively dmM and
d′ wM = dδwM . The female rates are defined analogously. Male and female
birth rates are set equal. In both cases, wildtype births occur in a given
household at rate

bZm F (Zm + Zw ), (1)

6
where
mM wM
Zm = (mF + wF (1 − v)ϕ) + (mF (1 − u) + wF (1 − v)ϕ) .
mM + wM mM + wM
(2)

Wolbachia-infected births occur in a given household at rate

bZw F (Zm + Zw ), (3)

where

Zw = vϕwF . (4)

These birth rates follow a similar formulation to Barlow et al. (2025a), except
here we account for male and female mosquitoes explicitly. In (1) and (3), b
is the maximum per capita (female) birth rate and F (Z) accounts for larval
density dependent mortality. Zm in the wildtype birth rate (1) is comprised
of two parts, as detailed in (2). The first term is the probability that a wild-
type male mates with a wildtype female, or mates with an infected female but
vertical transmission does not occur. The second term is the probability that
an infected male mates with a wildtype female and the offspring is viable, or
mates with an infected female and vertical transmission does not occur. The
offspring of a Wolbachia-infected male and a wildtype female are viable with
probability u. The fitness cost of Wolbachia-infection on the birth rate is
1 − ϕ. For the Wolbachia-infected birth rate (3), all mating interactions with
Wolbachia-infected female mosquitoes produce infected offspring if vertical
transmission is successful. The schematic in Figure 1 details all the differ-
ent mating interactions and outcomes that can occur between the male and
female wildtype and Wolbachia-infected mosquitoes via the CI effect. All
parameter values used throughout this study are detailed in Table 1.

2.1. The intraspecific larval density function

The function F (Zm + Zw ) represents the effects of intraspecific compe-
tition at the larval mosquito stage. We consider Zm + Zw here, instead of
the total adult mosquito population size mM + mF + wM + wF , in order
to account for the unhatched eggs produced from unviable births due to CI
pairings, which will have no impact on the competition among larvae for
resources. F (Zm + Zw ) is a monotonic decreasing function of Zm + Zw , with
F (0) = 1. Note that the birth rates are maximised when F (Zm + Zw ) = 1.

7
Table 1: Parameter definitions and values. Values except K, b, τ , ρ, γ and α are from
Hughes and Britton (2013) and re-scaled to mosquitoes per 100m2 , the approximate size of
a household dwelling. K is based on Madewell et al. (2019) and Barlow et al. (2025a); b is
calculated in Barlow et al. (2025a). The dispersal parameters τ , ρ, γ and α are estimated
in this study.

Parameter Definition Value

b Per capita (female) birth rate for wildtype mosquitoes 4.52d = 0.54 day−1

d Per capita death rate of wildtype mosquitoes 0.12 day−1

d′ = δd 0.12 day−1
Per capita death rate of Wolbachia-infected mosquitoes

b′ = ϕb Per capita (female) birth rate for Wolbachia-infected 0.52ϕ = 0.46 day−1
mosquitoes
u Cytoplasmic incompatibility. Probability of a 1
Wolbachia-infected male and uninfected female
producing inviable offspring
v Vertical transmission. Fraction of offspring of a 1
Wolbachia-infected female that are infected
ϕ 0.85
Reproductive cost of infection. See b′ above

δ 1
Mortality cost of infection. See d′ above

K Reproductive carrying capacity (of female mosquitoes) 30

k Larval competition parameter 0.3

h Larval competition parameter 0.19 × 100k = 0.76

τ Male mosquito household leave rate 0.15 (low), 0.33 (medium),

1 (high) day−1
ρ Male mosquito household arrival rate 0.15 (low), 0.33 (medium),
2 (high) day−1
γ Rate mosquitoes disperse from release site 0.15 (low), 0.33 (medium),
2 (high) day−1
α Rate female mosquitoes switch between households 0.03 (low), 0.1 (medium),
0.5 (high) day−1

8
Figure 1: Schematic of the CI effect depicting the possible reproductive outcomes in the
mating interactions between wildtype and Wolbachia-infected male and female mosquitoes.
The top row of circles represent the male (M) and female (F) mosquito in each possible
mating interaction. The circles below the arrows represent the offspring produced. Red
circles correspond to Wolbachia-infected mosquitoes and green circles correspond to wild-
type mosquitoes. The crossed grey outlined circles represent unviable offspring.

We discuss and implement two choices of the larval density function in Barlow
et al. (2025a). Here, we use
(
exp(−h(Zm + Zw )k ) mF + wF < K,
F (Zm + Zw ) = (5)
0 mF + wF ≥ K.

This is the heuristic function used in Hughes and Britton (2013); Dye (1984)
with a small modification. The birth rate is set to zero when the number
of female mosquitoes in the household reaches or exceeds the (female) re-
productive carrying capacity, K. This convention aligns with our previous
framework (Barlow et al., 2025a) and ensures the number of possible house-
hold states remains computationally tractable.

2.2. Accounting for movement and metapopulation effects

We consider a community of 100 households. In each household, mosquito
population dynamics occur by a stochastic birth-death process as described
above. In addition, we account for the movement of mosquitoes between
households. There are reported to be distinct behavioural differences be-
tween male and female Ae aegypti mosquitoes in terms of their mobility
(Juarez et al., 2020; Marcantonio et al., 2019; Trewin et al., 2020, 2021).
Male mosquitoes disperse large distances in the search of mates. In contrast,
female Aedes Aegypti tend to stay in areas with access to regular blood meals
and resting sites. In order to capture these different behaviours in our model
male mosquitoes leave their household at per capita rate ρ and enter a single
‘free-living’ pool. They move from this pool back into a household (chosen
at random) at rate τ . In contrast, female mosquitoes switch directly between
households at a low rate α. Initially we assumed female mosquitoes did not

9
move at all. However, we found that this approximation meant that house-
hold mosquito populations were unable to recover from extinction if their
last female inhabitant died.
The low-level female dispersal produces a rescue effect in the model. The
rescue effect is a phenomenon observed in metapopulation systems where the
probability of localised extinction, and ultimately metapopulation extinction,
is reduced by migration. Small subpopulations are subject to local extinction
due to stochastic fluctuations. A strong rescue effect is said to occur when
the arrival of female migrants boosts the birth rate in the population. A
weak rescue effect is said to occur when the arrival of migrants re-establishes
a previously extinct subpopulation. Both scenarios are possible under the
framework presented in this study. Metapopulation dynamics have been ex-
tensively documented in species residing in fragmented landscapes (Hanski
and Gaggiotti, 2004). The rescue effect is greatest when subpopulation dy-
namics are either uncorrelated or negatively correlated. Correlation in sub-
population dynamics can lead to metapopulation-wide extinction events due
to synchronisation (Keeling et al., 2004). The metapopulation as a whole,
as well as its individual component subpopulations, are often much more
stable and persist for longer than small populations in solitude (Hanski and
Gaggiotti, 2004).
It is possible to write down a Master Equation to describe the rate of
change of the household state probability using the single household approx-
imation (Holmes et al., 2022). We present this work in the Supplementary
Information. However, accounting for movement between households intro-
duces nonlinearities that render the Master Equation intractable. Hence, we
are unable to utilise many of the analytic techniques introduced in previous
work (Barlow et al., 2025a). Consequently, we use numerical simulation, via
the Gillespie SSA (Gillespie, 1977) in order to gain insight into the Wolbachia
invasion dynamics in a system composed of coupled household structures.

2.3. Release framework

We use our model to investigate the outcomes of a range of scenarios for
the release of Wolbachia-infected mosquitoes.
For community scale releases, we make a clear distinction between Wolbachia-
infected mosquitoes dispersing from the release site and male mosquitoes
moving between households and by introducing a separate mosquito popu-
lation class in the model, termed the ‘release’ population. The Wolbachia-
infected mosquitoes are added instantaneously to the free-living ‘release’ pop-

10
ulation and then disperse into households at per capita rate γ. Once they
enter a household, these mosquitoes become subject to the regular male and
female movement models described above.
Figure 2 shows a schematic representation of the model and release strat-
egy. The household habitats of the mosquitoes are denoted in brown. Each
household is characterised by its household state (mM , mF , wM , wF ), de-
noting the number of wildtype male mM (dark green), female mF (light
green) and Wolbachia-infected male wM (dark red), female wF (light red)
mosquitoes currently residing in the household. Female mosquitoes move
directly between households as indicated by the lighter blue arrows. Male
mosquitoes move from households into the free-living population (denoted
by the lower dashed circle of mosquitoes) and then into new households as
denoted by the darker blue arrows. Wolbachia-infected mosquitoes may be
released at a community scale or a household scale. The upper dashed circle
of infected (red) mosquitoes represents a community scale release.

2.4. Stochastic simulation

We solved our model by implementing Gillespie’s SSA in Julia. We used a
similar computational framework to Barlow et al. (2025b), employing dictio-
naries to keep track of the dynamic set of extant household states. We used
mutable structs to define the community of households. Each household
state was an instance of the struct, characterised by differing propensities
of the various household events. We defined the population of free-living
mosquitoes analogously using additional mutable structs. We distinguished
between mosquitoes released into the free-living pool that had not yet visited
a household and mosquitoes that had previously visited a household.
We considered communities composed of 100 households. Each household
was initialised with 8 male and 8 female wildtype mosquitoes, close to the
steady state value found for a single household in Barlow et al. (2025a).
For each realisation of the model the stochastic population dynamics were
simulated for 150 days from this starting point so that the wildtype household
mosquito populations approached the quasi-stationary distribution before
any Wolbachia-infected mosquitoes were released.
The code used to produce all results in this study can be found in the
GitHub repository https://github.com/ahb48/Wolbachia-movement-households.
In the Supplementary Material, we reproduce analytic results from our pre-
vious work (Barlow et al., 2025a) numerically using the SSA and investigate

11
the impact of different household population sizes and rates of dispersal be-
tween the households on the probability of a successful Wolbachia invasion
into a resident wildtype mosquito population.

2.5. Release scenarios

In Subsection 3.1, we investigate a single release of mixed sex Wolbachia-
infected mosquitoes into a community of N = 100 households. We con-
sider a community scale release with 400 male and 400 female Wolbachia-
infected mosquitoes introduced to the free-living ‘release’ pool. We also con-
sider a household scale release with 4 male and 4 female Wolbachia-infected
mosquitoes introduced directly to each household in the community. We com-
pare four release scenarios: community scale releases with low (γ = 0.15),
moderate (γ = 0.33), and high (γ = 2) release dispersal, and a household
scale release also as described above. We find that household scale releases
facilitate more rapid invasion of Wolbachia-infected mosquitoes, and that
community scale releases with higher dispersal correspond more closely with
household scale releases. In Subsection 3.2 we investigate single male-only
releases of a total of 800 mosquitoes at the community and household scales.
In Subsection 3.3 we explore the impact of varying the proportion of male
and female Wolbachia-infected mosquitoes in a release. We find that, due
to the CI effect, this ratio affects the invasion dynamics within the release
households, and the persistence of Wolbachia-infection in the metapopula-
tion.
In practice it will rarely be possible to control the wildtype mosquito
population with a single Wolbachia release event. So, in Subsection 3.4 and
the Supplementary Information we investigate the impact of releases carried
out at regular intervals in a single household. We consider the protection
this affords the release household, and the wider community.

3. Results
For each release scenario we computed 100 independent realisations of
the model. We derived several statistics from these ensembles to characterise
the invasion, establishment and persistence of Wolbachia-infection at the
household and community scales.
We calculated the mean and variance of the number of mosquitoes of
each type (male/female, wildtype/infected) in a household at a given time
point from the set of 10, 000 households formed from all 100 households in

12
Figure 2: Schematic of the model framework. Household habitats of the mosquitoes are de-
noted in brown. Each household is characterised by its household state (mM , mF , wM , wF ),
denoting the number of wildtype male mM (dark green), female mF (light green) and
Wolbachia-infected male wM (dark red), female wF (light red) mosquitoes currently resid-
ing in the household. Female mosquitoes move directly between households as indicated by
the lighter blue arrows. Male mosquitoes move from households into the free-living pop-
ulation (denoted by the lower dashed circle of mosquitoes) and then into new households,
as denoted by the darker blue arrows. Wolbachia-infected mosquitoes may be released
at a community scale or a household scale. The upper dashed circle of infected (red)
mosquitoes represents a community scale release.

13
all 100 realisations. The mean population size is an indicator of the extent to
which Wolbachia-infected mosquitoes have replaced wildtypes in the entire
metapopulation. The variance of the population size is an indicator of the
uniformity of the Wolbachia invasion across the metapopulation.
We measured the persistence time of Wolbachia infection in the entire
community, or in a given household, as the proportion of the 850 day simula-
tion period following the release on day 150 for which at least one Wolbachia-
infected mosquito was present. For persistence within a given household, in
each realisation of the model, the state of a single household in the commu-
nity was monitored for the full duration of the simulation to get the duration
of infection. The outcome was then averaged over 100 realisations.

3.1. Community versus household scale single releases

Figure 3a shows the mean number of Wolbachia-infected mosquitoes in
a household over time for four release scenarios: household scale release and
community scale release with low, moderate and high dispersal rates from
the release site. Figure 3b shows the corresponding variance and Figure
3c) decomposes this variance into the male and female Wolbachia-infected
subpopulations in the household. Figure 4 shows how the mean number
of households that contain no wildtype mosquitoes varies over time. Table
2 shows the mean persistence time of Wolbachia-infection in a household
mosquito population.
The mean population trajectories reach a quasi-steady state after around
60 weeks. In general, lower dispersal leads to smaller Wolbachia-infected
household mosquito populations with higher variance. When dispersal is
low, there is a higher chance of mortality before reaching a household. Fur-
thermore, for each household, there is less momentum to drive the initial
invasion into the resident wildtype population and so more variability in the
outcome. The variance in the female subpopulation is more than double
that of the male subpopulation. This is because females move less frequently
between households than males after the initial release dispersal. All four re-
lease scenarios lead to communities in which there are no wildtype mosquitoes
in any household after 45 weeks, although the Wolbachia-infected household
populations may not yet have reached quasi-steady state. Lower dispersal is
associated with slower eradication of the wildtype mosquito population and
shorter mean persistence times of Wolbachia infections in household popula-
tions.

14
 Release scenario Proportion of time (days)
community, γ = 0.15 89.3%
community, γ = 0.33 96.4%
community, γ = 2.0 99.6%
household 99.7%
Table 2: Proportion of time for which there is at least one Wolbachia-infected mosquito in
the household over the period between day 150 and day 1000. For each realisation of the
model, the state of a single household was monitored for the full duration of the simulation
to get the duration of infection. The results shown are mean values calculated from 100
realisations. For model configuration and parameterisation see the caption of Figure 3.

The invasion dynamics arising from a household scale release are very sim-
ilar to those of a community scale release with high dispersal (γ = 2). So, in
the absence of rapid dispersal of community-released mosquitoes, household
scale releases may support the establishment of substantial and persistent
Wolbachia-infected household mosquito populations.

3.2. Male only single release

Here we explore the same four release scenarios as described in Section 2.3,
except each release is made up of male mosquitoes only (800 males released
in total).
Figure 5 shows the mean number of wildtype mosquitoes per household
over time. Households reached a quasi-steady state of around 14 wildtype
mosquitoes around 7 weeks after the simulation was initiated. The infected
males were introduced after 150 days. The wildtype population was immedi-
ately suppressed, but only temporarily. A household scale release was most
effective, decreasing the mean number of wildtype mosquitoes in a household
to just below 11. A community scale release with high dispersal produced
a similar result. Community scale releases with moderate and low disper-
sal brought the mean wildtype household population down to around 12
mosquitoes. Under all the release scenarios, the wildtype mosquito popula-
tions recovered to the original quasi-steady state within around 15 weeks of
the release. We further observed that, on average, male Wolbachia-infected
mosquitoes became extinct in the household less than 2 weeks after the re-
lease (result is not shown here).
Clearly, regular releases of male Wolbachia-infected mosquitoes are needed
in order for this strategy to be effective. A single male-only release strategy is

15
 10.0

Mean number of Wolbachia-infected

mosquitoes (per household)
7.5

5.0

2.5 = 0.15
= 0.33
= 2.0
household release
0.0
0 30 60 90 120
Time (weeks)
(a)

25 12.5
Variance of Wolbachia-infected

Variance of Wolbachia-infected
mosquitoes per household

mosquitoes per household

20 10.0 females

15 7.5

10 5.0
= 0.15 males = 0.15
5 = 0.33 2.5 = 0.33
= 2.0 = 2.0
household release household release
0 0.0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time (weeks) Time (Weeks)
(b) (c)

Figure 3: Mean and variance of the number of Wolbachia-infected mosquitoes per house-
hold over time (weeks) for different release scenarios. (a) Mean for male and female
mosquitoes combined. (b) Variance for male and female mosquitoes combined. (c) Vari-
ance of male and female mosquitoes separately. Statistics were calculated from 100 model
realisations each made up of 100 households. A single release of Wolbachia-infected
mosquitoes was carried out at time 150 days. The coloured curves denote community
scale releases of 800 mosquitoes (400 male, 400 female) for low (blue), moderate (red) and
high (green) dispersal release rates γ. The black dashed curve denotes a household scale
release of 8 mosquitoes (4 male and 4 female) in each household. All other parameter
values can be found in Table 1, with the intermediate values for the non-release dispersal
parameters.

16
 100
= 0.15
= 0.33
=2

Mean number of wildtype-free

75 Household releases

households
50

25

0
0 10 20 30 40 50 60
Time (weeks)
Figure 4: Mean number of households containing no wildtype mosquitoes over time (weeks)
for different release scenarios. For model configuration and parameterisation see Figure 3.

much less effective than a single mixed sex release strategy as a consequence
of the CI effect. When a Wolbachia-infected male enters a wildtype house-
hold it is unable to produce viable (and infected) offspring with the wildtype
females. Consequently, infected males can dilute wildtype reproduction in
the household, but cannot drive invasion of Wolbachia infection.

3.3. The impact of mosquito sex ratios in household scale releases

Figure 6 shows the results of a single household scale release of f female
and m male wildtypes for m, f = 0, 1, 2, . . . , 10 in half (50) of the households.
Figure 6a shows that wildtype extinction of the whole population of house-
holds is only possible if at least one female Wolbachia-infected mosquito is
released in each household. As discussed in Subsection 3.2, this is because
under the CI effect, an infected male is unable to produce Wolbachia-infected
offspring with a wildtype female. Releasing 3 Wolbachia-infected females in
each household is sufficient for metapopulation-wide Wolbachia invasion to
occur with a high probability. The number of infected males released into the
household only has a noticeable impact when the release includes less than 4
infected females per household. In this case, higher numbers of Wolbachia-
infected males increase the probability of wildtype extinction because infected
females are more likely to produce infected offspring before dying or switch-
ing households. Lowering the rate at which female mosquitoes disperse does
not qualitatively impact this result.

17
 16

15

mosquitoes (per household)

Mean number of wildtype
14

13

12 = 0.15
= 0.33
= 2.0
11 household release
0 30 60 90 120
Time (weeks)
Figure 5: Mean number of wildtype mosquitoes per household over time (weeks) with
a single male-only release of Wolbachia-infected mosquitoes at 150 days. The coloured
curves denote community wide releases of 800 male mosquitoes for low (blue), moderate
(red) and high (green) release dispersal rates γ. The black dashed curve represents a
household release strategy where 8 male Wolbachia-infected mosquitoes are released in
each household. All other parameter values can be found in Table 1, with the intermediate
values for the non-release dispersal parameters.

As indicated in Figure 6b, the persistence of Wolbachia-infection improves

with the number of female Wolbachia-infected mosquitoes released. When at
least 5 females are released into each household, Wolbachia-infection persists
in the metapopulation for the majority of the simulation. The persistence
time is insensitive to the number of males released.

3.4. Regular releases into a single household

Figure 7 shows results from simulations in which Wolbachia-infected mosquitoes
are regularly released in a single household. Figures 7a and 7b show the pro-
portion of households (10, 000 households composed of 100 households in
each of 100 realisations) for which the release and non-release households are
wildtype-free at time 1000 days. If Wolbachia invasion is successful in the
release household then it almost always successfully invades the non-release
households as well. Invasion with a probability greater than 0.8 requires at
least 5 female Wolbachia-infected mosquitoes to be released at least every
10 days. If 10 infected females are released, an invasion probability greater
than 0.5 requires a release every 25 days or less. For larger time intervals be-
tween releases or smaller release populations, the probability of a successful
Wolbachia invasion is small. We hypothesise that smaller and less frequent

18
 (a) (b)

Figure 6: Outcomes after 1000 days following a single release of f female and m male
Wolbachia-infected mosquitoes into 50% of households at 150 days. All households were
initialised with 8 male and 8 female wildtype mosquitoes. All parameters can be found in
Table 1, where we take intermediate values for the dispersal parameters. (a) Probability
of wildtype extinction in the entire metapopulation. (b) Proportion of time, after day 150,
for which Wolbachia infection persists in the metapopulation.

releases will be sufficient if the proportion of households participating in re-

leases is higher.
Our results show that regular releases of infected mosquitoes in a single
household can sustain Wolbachia-infection in an entire community. Closer in-
spection of the results reveals that, although the Wolbachia-infected mosquitoes
spread to all non-releasing households, wildtype replacement in these house-
holds can take several years. In the release household wildtype replacement
occurs within a few weeks. Wolbachia-infected mosquitoes suffer a fitness
cost in terms of reduced reproduction. The CI effect can compensate for this
cost, but only if the infected proportion of the household mosquito popula-
tion is sufficiently high. In the release household, repeated introduction of
significant numbers of infected mosquitoes generate sufficient momentum for
invasion to occur with high probability. In the non-release households, inva-
sions are typically driven by very small numbers of infected mosquitoes, and
the weak momentum means that multiple invasion attempts may be required
before successful establishment. When fewer female (and male) mosquitoes
are released, or releases occur less frequently, it takes much longer to build
sufficient momentum for successful invasion, and it may take several years
for the household to reach a quasi-steady state.

19
 In Figures 7e and 7f we compare the proportion of time for which Wolbachia-
infection (i.e. at least one infected mosquito) is present in the release and
non-release households. More frequent releases support longer persistence.
In all cases, Wolbachia-infection persists for considerably longer in the re-
lease households than in households in the wider community, indicating the
potential for a localised protective effect.
Smaller time intervals between releases and larger releases of infected
females produce larger numbers of Wolbachia-infected mosquitoes in the re-
lease household at time 1000 days (see Figure 7c). When releases are carried
out every 10 days and at least 10 females are released each time, the non-
release households reach a steady state of 10 wildtype mosquitoes, whereas
the continual releases push the Wolbachia-infected population in the release
household above the steady state. When releases are carried out more than
25 days apart, the mean number of Wolbachia-infected mosquitoes in a non-
release household becomes insignificant, indicating that infection is unlikely
to spread from the release household (see Figure 7d).

4. Discussion
In this study, we developed a model for the release and establishment
of Wolbachia-infected mosquitoes at a household scale. We were motivated
by the observation that Ae aegypti mosquitoes often live in urban settings,
in and around the dwellings of the people that they bite (Dzul-Manzanilla
et al., 2017). We utilised stochastic simulation (Gillespie, 1977) to keep track
of a dynamic set of mosquito household populations, in addition to free-living
mosquito populations arising from community scale releases and mosquitoes
migrating between household habitats. Our primary aim was to investigate
the replacement of resident wildtype populations by introduced Wolbachia-
infected populations occurring stochastically at small population scales.
We found that for mixed sex releases, releasing Wolbachia-infected mosquito
populations at a household scale generally leads to more successful invasion
and a higher quasi-steady state Wolbachia-infected population in the house-
hold than performing a community wide release of the same total number of
mosquitoes.
Highly localised releases are effective because the Wolbachia-infected mosquitoes
suffer a fitness cost due to infection but a frequency-dependent fitness bene-
fit due to cytoplasmic incompatibility — the offspring of infected males and

20
 Release household Non-release households
1.0 Sex ratio (f,m) 1.0 Sex ratio (f,m)
Probability of wildtype-free

Probability of wildtype-free
(1, 1) (1, 1)
0.8 (3, 1)
(3, 3) 0.8 (3, 1)
(3, 3)
(3, 5) (3, 5)
household

household
0.6 (3, 10) 0.6 (3, 10)
(5, 1) (5, 1)
(5, 5) (5, 5)
0.4 (5, 10) 0.4 (5, 10)
(10, 1) (10, 1)
0.2 (10, 5)
(10, 10)
0.2 (10, 5)
(10, 10)
0.0 0.0
10 25 50 75 100 10 25 50 75 100
Time between releases (days) Time between releases (days)
(a) (b)
Mean number of Wolbachia-infected

Mean number of Wolbachia-infected

Release household Non-release households
35 10
mosquitoes (per household)

mosquitoes (per household)

Sex ratio (f,m) Sex ratio (f,m)
30 (1, 1) (1, 1)
25
(3, 1)
(3, 3) 8 (3, 1)
(3, 3)
(3, 5) (3, 5)
20 (3, 10) 6 (3, 10)
(5, 1) (5, 1)
15 (5, 5)
4 (5, 5)
(5, 10) (5, 10)
10 (10, 1) (10, 1)
5
(10, 5)
(10, 10)
2 (10, 5)
(10, 10)
0 0
10 25 50 75 100 10 25 50 75 100
Time between releases (days) Time between releases (days)
(c) (d)
Proportion of time household contains

Proportion of time household contains

1.0 Release household 1.0 Non-release households

Wolbachia-infected mosquitoes

Wolbachia-infected mosquitoes

Sex ratio (f,m)

(1, 1)
0.8 0.8 (3, 1)
(3, 3)
(3, 5)
0.6 0.6 (3, 10)
(5, 1)
(5, 5)
0.4 0.4 (5, 10)
(10, 1)
(10, 5)
0.2 0.2 (10, 10)

0.0 0.0
10 25 50 75 100 10 25 50 75 100
Time between releases (days) Time between releases (days)
(e) (f)

Figure 7: Outcomes of regular releases of Wolbachia-infected mosquitoes in a single house-

hold. All households were initialised with 8 male and 8 female wildtype mosquitoes.
Starting at day 150, Wolbachia-infected mosquitoes were introduced into the same single
household with the frequency shown on the horizontal axis. Line colours indicate the
numbers of male m and female f infected mosquitoes released (f, m). (a), (b) Probability
a household is wildtype-free after 1000 days; (c), (d) Mean number of Wolbachia-infected
mosquitoes in a household after 1000 days; (e), (f) Proportion of time after day 150 for
which Wolbachia-infection is present in a household. Plots (a), (c), and (e) correspond to
the household where releases occur; plots (b), (d), and (f) correspond to the other house-
holds in the community. For plots (e) and (f), the proportion of time Wolbachia-infection
is present in the household is the mean derived from monitoring a single release household
and a single non-release household in each of 100 model realisations. For all other plots
we average over a total of 10, 000 households (100 households for each 100 model realisa-
tions). All parameters can be found in Table 1, where we take intermediate values for the
dispersal parameters. 21
wildtype females are not viable. Therefore, the probability of Wolbachia-
infected mosquitoes successfully invading a household is higher when the
initial infected population is larger. Household scale release provides lo-
calised momentum for invasion. In general, community scale releases reduce
the invasion momentum because the mosquitoes disperse gradually and enter
households sporadically. Community scale releases with very high dispersal
rates can produce invasion dynamics comparable to household scale releases.
However, we believe such a dispersal rate is improbable in urban environ-
ments.
Community scale and household scale release strategies converge as the
rate of initial dispersal into the households (γ) becomes large. We performed
a sensitivity analysis for the single mixed sex release scenario, varying the
dispersal rates between low, moderate and high values in all possible combi-
nations of our four dispersal parameters. Our full set of dispersal parameters
are (1) the rate at which released Wolbachia-infected mosquitoes disperse
into the households (γ); (2) the rate at which male mosquitoes arrive at
households from the free-living population (ρ); (3) the rate at which male
mosquitoes leave households for the free-living population (τ ) and (4) the rate
at which female mosquitoes switch between households (α). The simulation
data produced from our sensitivity analysis can be found in the GitHub repos-
itory https://github.com/ahb48/Wolbachia-movement-households. We
observed no qualitative changes in the overall pattern of results for any of
the dispersal rate parameter sets we considered.
We also considered a male-only release of Wolbachia-infected mosquitoes
with the aim of suppressing the wildtype population in the household. Al-
though we concluded that the household scale release was once again most ef-
fective at pushing the wildtype population in the household below the steady
state value, under all release scenarios the wildtype population quickly re-
bounded to its original steady state value following a single release event.
This result suggests that if long term suppression is required, numerous re-
leases may need to be carried out in quick succession until the wildtype
population has become extinct both in the household and in the surrounding
community. Alternatively, releases may be carried out on a regular basis in
order to maintain the wildtype population at a lower level that reduces or
eliminates the risk of vector-borne disease circulation in the community.
The goal of the control strategy differs between mixed-sex and male-only
releases and on the strain of Wolbachia used. Under the mixed-sex release,
in general, the aim is to replace the wildtype population with a Wolbachia-

22
infected population. This is because mosquitoes infected with certain strains
of Wolbachia are less competent at passing on dengue infection to humans.
However, the female mosquitoes will still take blood meals from the inhab-
itants of the household and some transmission is still possible. Under a
male-only release, the goal is to suppress the total mosquito population via
the infected males diluting the reproduction of mosquitoes through the CI
effect. This leads to fewer (potentially no) mosquitoes biting human in-
habitants and consequently lower transmission risk. It may be possible to
achieve something in between these two strategies by varying the sex ratio of
the Wolbachia-infected mosquitoes released. This could yield new strategies
that minimise resource expenditure and disease risk.
We explored this premise and found that releasing very small numbers of
infected female mosquitoes in 50% of households leads to widespread inva-
sion of the metapopulation with high probability. The number of number of
infected male mosquitoes released per household has little impact on the in-
vasion dynamics. Therefore, in general, mixed-sex releases offer no advantage
over female-only releases.
A single release event is unlikely to be sufficient for long term control
of the wildtype mosquito population, particularly if CI is imperfect or wild-
type mosquitoes re-invade from regional populations. So we adapted our
simulation framework to account for household scale releases occurring at
regular intervals. We determined that even a single household performing
releases of at least 10 female mosquitoes at least every 10 days, can even-
tually protect a whole community. However, it can take several years to
achieve wildtype replacement in non-release households in this way whereas
wildtype replacement in the release household is rapid. When releases are
carried out less frequently or with fewer female mosquitoes, it seems possi-
ble that some bistable quasi-steady state may be reached in the household
metapopulation. Similar effects, where some subpopulations in the metapop-
ulation reach a certain steady state solution while other subpopulations find a
different steady state, have been observed in deterministic ecological models
(Vilk and Assaf, 2020). We were not able to identify any model realisations
in which some households reached a wildtype-only quasi-steady state while
others reached a Wolbachia-only quasi-steady state.
In a previous study (Barlow et al., 2025a), we used analytical methods
from probability theory to investigate the outcome of Wolbachia-infected
mosquitoes invading a single household containing a resident wildtype pop-
ulation. Here we have extended the model to a community of households

23
connected by the movement of mosquitoes. This dispersal introduced non-
linearities into the model that made it necessary to move towards numerical
approaches. For single households, we found good agreement between our
analytic and numerical frameworks. There is further discussion on this in
the Supplementary Information. There is, however, a small discrepancy be-
tween the wildtype quasi-steady state number of mosquitoes in our current
model and the corresponding single household steady state from Barlow et al.
(2025a). We believe this is a consequence of the mosquito birth rate being
influenced by the sex ratio when males are modelled explicitly. The dispersal
of male mosquitoes further alters the household steady state since at any
time a number of male mosquitoes are outside of the household, in the free-
living population. We do not think this discrepancy is likely to have any
qualitative impact on the model dynamics.
The numerical framework we have implemented has the additional ad-
vantage of accommodating a wide range of variation in community structure
and release strategies. This flexibility can be exploited to further inform
Wolbachia-based control strategies and field trials in dengue prevalent areas.
In future work we aim to explore a wider range of release strategies. For
instance, adaptive strategies could monitor the mosquito population sizes
in the community and the household and regularly adjust the numbers of
male and female mosquitoes released. Alternatively, it could be desirable to
prioritise highly infested households or households with clinically vulnerable
residents. These studies could reveal more efficient and cost effective control
strategies.
We acknowledge several limitations and simplifying assumptions in our
framework. The size of the mosquito population in a household is restricted.
Free-living mosquitoes can only enter a household that contains less than 2K
mosquitoes in total. Mosquitoes are only born into a household if there are
less than K females residing there. We place a restriction on household size
in order to keep the number of possible household states computationally
manageable. Our bound is well within the range suggested by empirical
studies (Madewell et al., 2019). The birth rate restriction originates from our
framework in Barlow et al. (2025a) where we model the female population
only. This bound is retained here verbatim for consistency. A consequence
of this is that it is possible for the household to move above the household
reproductive carrying capacity, 2K, when there are less than K females but
2K or more total mosquitoes in the household. This situation rarely occurs
and we do not believe it has a significant effect on the results.

24
 We introduce a rescue effect into the metapopulation by allowing fe-
male mosquitoes to switch at a low rate between households, whereas male
mosquitoes move from a household to a ‘free-living’ population, and then into
another household. Our framework is set up in this way to capture empirical
observations of females mosquitoes dispersing over much smaller distances
than males. As more data becomes available, it will be beneficial to explore
the contrasting dispersal behaviours of male and female mosquitoes in more
detail.
We assume vertical transmission and the CI effect are perfect, so that
Wolbachia-infected females produce infected offspring with probability v = 1
and 100% of offspring produced by infected males and wildtype females are
unviable. Empirical studies (Hoffmann et al., 2011; Xi et al., 2005; McMeni-
man et al., 2009) suggest this is a reasonable assumption. In our previous
study (Barlow et al., 2025a), we explored the impact of imperfect vertical
transmission in Wolbachia-infected female mosquitoes. We found it did not
significantly affect the dynamics of probability of Wolbachia invasion as long
as 0.9 ≤ v < 1.
We defined a larval density dependence function acting on the birth rate,
which describes the effect of intraspecific competition for resources such as
food and space. The function is slightly adjusted in comparison to the larval
density function used in Barlow et al. (2025a) in order to account for unvi-
able births produced via the CI effect which will not contribute towards the
competition. The function is in line with that used in Hughes and Britton
(2013); Dye (1984) and many other modelling studies. The model is heuris-
tic, however data was originally used to estimate the model parameters in
(Dye, 1984) under the original scaling. Were more data to become available,
it could be beneficial to revisit the model.
In conclusion, Wolbachia-infected mosquito release is a promising ap-
proach to mosquito control and consequently management of vector-borne
infections such as dengue. Aedes aegypti mosquitoes are known to reside in
urban areas, in and around the same dwellings as the people that they bite.
In this study we accounted for stochastic effects in the invasion and estab-
lishment of Wolbachia-infected mosquitoes at a household scale and found
that household scale releases are a promising avenue for efficient mosquito
control.

25
5. Supplementary information
All code used to produce the results in this study can be found at https:
//github.com/ahb48/Wolbachia-movement-households. Supplementary
Information can be found in a supporting document online.

6. Acknowledgements
This work was supported by a scholarship from the EPSRC Centre for
Doctoral Training in Statistical Applied Mathematics at Bath (SAMBa),
under the project EP/S022945/1.
For the purpose of open access, the author has applied a Creative Com-
mons Attribution (CC-BY) license to any Author Accepted Manuscript ver-
sion arising. No new data was created during the study.

References
Barlow, A., Penington, S., Adams, B., 2025a. Analysis of a household-scale
model for the invasion of wolbachia into a resident mosquito population.
arXiv preprint arXiv:2502.12833 .

Barlow, A., Penington, S., Adams, B., 2025b. Epidemiological dynamics

in populations structured by neighbourhoods and households. Bulletin of
Mathematical Biology 87, 50.

Bhatt, S., Gething, P.W., Brady, O.J., Messina, J.P., Farlow, A.W., Moyes,
C.L., Drake, J.M., Brownstein, J.S., Hoen, A.G., Sankoh, O., et al., 2013.
The global distribution and burden of dengue. Nature 496, 504–507.

CDC, 2022. Mosquitoes with Wolbachia for reducing numbers of

Aedes aegypti mosquitoes. https://www.cdc.gov/mosquitoes/
mosquito-control/community/emerging-methods/wolbachia.html.
Accessed: 04/08/22.

Dorigatti, I., McCormack, C., Nedjati-Gilani, G., Ferguson, N.M., 2018. Us-
ing Wolbachia for dengue control: insights from modelling. Trends para-
sitol. 34, 102–113.

26
Dos Santos, G.R., Durovni, B., Saraceni, V., Riback, T.I.S., Pinto, S.B.,
Anders, K.L., Moreira, L.A., Salje, H., 2022. Estimating the effect of
the wmel release programme on the incidence of dengue and chikungunya
in rio de janeiro, brazil: a spatiotemporal modelling study. The Lancet
Infectious Diseases 22, 1587–1595.

Dufault, S.M., Tanamas, S.K., Indriani, C., Utarini, A., Ahmad, R.A., Jew-
ell, N.P., Simmons, C.P., Anders, K.L., 2022. Disruption of spatiotemporal
clustering in dengue cases by wMel Wolbachia in Yogyakarta, Indonesia.
Sci rep. 12, 1–11.

Dye, C., 1984. Models for the population dynamics of the yellow fever
mosquito, aedes aegypti. The Journal of Animal Ecology , 247–268.

Dye, D., Cain, J.W., 2024. Efficacy of wolbachia-based mosquito control:

Predictions of a spatially discrete mathematical model. Plos one 19,
e0297964.

Dzul-Manzanilla, F., Ibarra-López, J., Bibiano Marín, W., Martini-Jaimes,

A., Leyva, J.T., Correa-Morales, F., Huerta, H., Manrique-Saide, P.,
Vazquez-Prokopec, G.M., 2017. Indoor resting behavior of Aedes aegypti
(diptera: Culicidae) in Acapulco, Mexico. Journal of medical entomology
54, 501–504.

Gillespie, D.T., 1977. Exact stochastic simulation of coupled chemical reac-

tions. The journal of physical chemistry 81, 2340–2361.

Hancock, P.A., Ritchie, S.A., Koenraadt, C.J., Scott, T.W., Hoffmann, A.A.,
Godfray, H.C.J., 2019. Predicting the spatial dynamics of wolbachia in-
fections in Aedes aegypti arbovirus vector populations in heterogeneous
landscapes. Journal of Applied Ecology 56, 1674–1686.

Hanski, I.A., Gaggiotti, O.E., 2004. Ecology, genetics and evolution of

metapopulations.

Hasan, S., Jamdar, S.F., Alalowi, M., Al Beaiji, S.M.A.A., 2016. Dengue
virus: A global human threat: Review of literature. Journal of Interna-
tional Society of Preventive & Community Dentistry 6, 1.

Hoffmann, A.A., Goundar, A., Long, S., Johnson, P.H., Ritchie, S.A., 2014.
Invasion of Wolbachia at the residential block level is associated with local

27
 abundance of Stegomyia aegypti, yellow fever mosquito, populations and
property attributes. Med vet entomol. 28, 90–97.

Hoffmann, A.A., Montgomery, B., Popovici, J., Iturbe-Ormaetxe, I., John-

son, P., Muzzi, F., Greenfield, M., Durkan, M., Leong, Y., Dong, Y., et al.,
2011. Successful establishment of Wolbachia in Aedes populations to sup-
press dengue transmission. Nature 476, 454–457.

Holmes, A., Tildesley, M., Dyson, L., 2022. Approximating steady state
distributions for household structured epidemic models. Journal of Theo-
retical Biology 534, 110974.

Hughes, H., Britton, N.F., 2013. Modelling the use of Wolbachia to control
dengue fever transmission. Bull math biol. 75, 796–818.

Juarez, J.G., Garcia-Luna, S., Chaves, L.F., Carbajal, E., Valdez, E., Avila,
C., Tang, W., Martin, E., Barrera, R., Hemme, R.R., et al., 2020. Dispersal
of female and male Aedes aegypti from discarded container habitats using a
stable isotope mark-capture study design in south texas. Scientific Reports
10, 1–12.

Keeling, M.J., Bjørnstad, O.N., Grenfell, B.T., 2004. Metapopulation dy-

namics of infectious diseases.

Madewell, Z.J., Sosa, S., Brouwer, K.C., Juárez, J.G., Romero, C., Lenhart,
A., Cordón-Rosales, C., 2019. Associations between household envi-
ronmental factors and immature mosquito abundance in quetzaltenango,
guatemala. BMC Public Health 19, 1–11.

Marcantonio, M., Reyes, T., Barker, C.M., 2019. Quantifying Aedes aegypti
dispersal in space and time: a modeling approach. Ecosphere 10, e02977.

McMeniman, C.J., Lane, R.V., Cass, B.N., Fong, A.W., Sidhu, M., Wang,
Y.F., O’Neill, S.L., 2009. Stable introduction of a life-shortening Wolbachia
infection into the mosquito Aedes aegypti. Science 323, 141–144.

Ndii, M.Z., Hickson, R.I., Allingham, D., Mercer, G., 2015. Modelling the
transmission dynamics of dengue in the presence of wolbachia. Mathemat-
ical biosciences 262, 157–166.

28
Nguyen, T.H., Nguyen, H.L., Nguyen, T.Y., Vu, S.N., Tran, N.D., Le, T.,
Vien, Q.M., Bui, T., Le, H.T., Kutcher, S., et al., 2015. Field evaluation
of the establishment potential of w Melpop Wolbachia in Australia and
Vietnam for dengue control. Parasites & vectors 8, 1–14.
O’Neill, S.L., Ryan, P.A., Turley, A.P., Wilson, G., Retzki, K., Iturbe-
Ormaetxe, I., Dong, Y., Kenny, N., Paton, C.J., Ritchie, S.A., et al., 2018.
Scaled deployment of wolbachia to protect the community from dengue
and other aedes transmitted arboviruses. Gates open research 2.
Pagendam, D., Trewin, B., Snoad, N., Ritchie, S., Hoffmann, A., Staunton,
K., Paton, C., Beebe, N., 2020. Modelling the wolbachia incompatible
insect technique: strategies for effective mosquito population elimination.
BMC biology 18, 1–13.
Ross, P.A., 2021. Designing effective wolbachia release programs for mosquito
and arbovirus control. Acta tropica 222, 106045.
Strugarek, M., Bossin, H., Dumont, Y., 2019. On the use of the sterile insect
release technique to reduce or eliminate mosquito populations. Appl Math
Model. 68, 443–470.
Trewin, B.J., Pagendam, D.E., Johnson, B.J., Paton, C., Snoad, N., Ritchie,
S.A., Staunton, K.M., White, B.J., Mitchell, S., Beebe, N.W., 2021. Mark-
release-recapture of male aedes aegypti (diptera: Culicidae): Use of rho-
damine b to estimate movement, mating and population parameters in
preparation for an incompatible male program. PLOS Neglected Tropical
Diseases 15, e0009357.
Trewin, B.J., Pagendam, D.E., Zalucki, M.P., Darbro, J.M., Devine, G.J.,
Jansen, C.C., Schellhorn, N.A., 2020. Urban landscape features influ-
ence the movement and distribution of the australian container-inhabiting
mosquito vectors Aedes aegypti (diptera: Culicidae) and Aedes notoscrip-
tus (diptera: Culicidae). Journal of medical entomology 57, 443–453.
Turelli, M., Barton, N.H., 2017. Deploying dengue-suppressing wolbachia:
Robust models predict slow but effective spatial spread in aedes aegypti.
Theoretical population biology 115, 45–60.
Vilk, O., Assaf, M., 2020. Extinction risk of a metapopulation under bistable
local dynamics. Physical Review E 101, 012135.

29
Walker, T., Johnson, P., Moreira, L., Iturbe-Ormaetxe, I., Frentiu, F., Mc-
Meniman, C., Leong, Y.S., Dong, Y., Axford, J., Kriesner, P., et al., 2011.
The w Mel Wolbachia strain blocks dengue and invades caged Aedes ae-
gypti populations. Nature 476, 450–453.

WHO, . Dengue and severe dengue. URL: https://www.who.

int/news-room/fact-sheets/detail/dengue-and-severe-dengue. ac-
cessed: 29/08/2023.

WHO, . Vaccines and immunization: Dengue. URL: https://www.who.int/

news-room/questions-and-answers/item/dengue-vaccines. accessed:
29/08/2023.

Xi, Z., Khoo, C.C., Dobson, S.L., 2005. Wolbachia establishment and inva-
sion in an aedes aegypti laboratory population. Science 310, 326–328.

Zeng, Z., Zhan, J., Chen, L., Chen, H., Cheng, S., 2021. Global, regional, and
national dengue burden from 1990 to 2017: A systematic analysis based
on the global burden of disease study 2017. EClinicalMedicine 32.

30