Modeling the potential of wAu-Wolbachia strain invasion in mosquitoes to control Aedes-borne arboviral infections.

Arboviral infections such as dengue, Zika and chikungunya are fast spreading diseases that pose significant health problems globally. In order to control these infections, an intracellular bacterium called Wolbachia has been introduced into wild-type mosquito populations in the hopes of replacing the vector transmitting agent, Aedes aegypti with one that is incapable of transmission. In this study, we developed a Wolbachia transmission model for the novel wAu strain which possesses several favourable traits (e.g., enhanced viral blockage and maintenance at higher temperature) but not cyctoplasmic incompatibility (CI)-when a Wolbachia-infected male mosquito mates with an uninfected female mosquito, producing no viable offspring. This model describes the competitive dynamics between wAu-Wolbachia-infected and uninfected mosquitoes and the role of imperfect maternal transmission. By analysing the system via computing the basic reproduction number(s) and stability properties, the potential of the wAu strain as a viable strategy to control arboviral infections is established. The results of this work show that enhanced maintenance of Wolbachia infection at higher temperatures can overcome the lack of CI induction to support wAu-Wolbachia infected mosquito invasion. This study will support future arboviral control programs, that rely on the introduction of new Wolbachia variants.

Introduction

Arthropod-borne viruses, or arboviruses, are viruses that are transmitted via blood feeding arthropods[1]. Arboviral infections such as dengue, Zika and chikungunya are fast spreading diseases that pose significant health problems globally[2-5]. These viral infections, in particular dengue, are transmitted mainly by Aedes aegypti and sometimes by Aedes albopictus (Asian Tiger) female mosquitoes when taking a blood meal from the host[6,7]. Approximately 390 million dengue infections are estimated to occur worldwide annually, putting 40% of the total human population at risk[8]. Dengue infection is the most geographically wide-spread of the arboviral infections[3,8]. It has different severity levels which are classified according to disease progression from dengue without warning signs to dengue with warning signs and then severe dengue[9]. Clinical manifestation includes sudden high-grade fever, headache, nausea, arthralgia, eye pain, muscle ache and rash in some cases[10]. Presently, there is no specific universal treatment for dengue infections: the vaccine envelopment targets young populations; the efficacy of the only vaccine licensed depends on prior immunity to at least one serotype of dengue; and it provides heterogeneous protection against the different serotypes[11,12].

Other arboviral infections such as Zika, chikungunya and yellow fever are also of global health concern[13]. These arboviral infections have occurred simultaneously with dengue[13,14]. Some of these infections share many similar clinical manifestations with dengue infection and also allow arboviral coinfection such as dengue and chikungunya[15], chikungunya and Zika[16] and yellow fever and chikungunya[17]. Although, there are no specific treatments for Zika and chikungunya viral infections, these infections can be managed by supportive treatment of symptomatic individuals and adequate rest. This treatment includes fluid intake and administering drugs such as acetaminophen to suppress pain and fever[18,19]. However the prevention strategy for yellow fever infection is available i.e. vaccination[20,21].

To control these infections, an intracellular bacterium called Wolbachia can be used to suppress transmission in arthropods such as mosquitoes and flies[22-25]. Wolbachia infection inhibits arboviral transmission in mosquitoes via four mechanisms: immune priming—preactivation of the mosquito immune system; induction of the phenoloxidase cascade—triggers immune response to viruses; competition of intracellular resources—inducing authophagy; and induction of microRNA-dependent immune pathways—essential for gene regulation and stability, immune defense, ageing and organ differentiation[26]. This endosymbiotic bacterium which exists naturally in more than 50% of all insect species can be found within the cytoplasm of the cells of their hosts[25,27,28]. Whilst Wolbachia is not naturally present in Aedes aegypti, it can be introduced via stable transinfections using microinjections[29,30].

The Wolbachia-based control strategy is carried out by infecting mosquitoes with a strain of Wolbachia and then releasing them into wild mosquito populations in the hopes of replacing the vector transmitting agent Aedes aegypti with one that is incapable of transmission[29-31]. Infecting an Aedes mosquito with Wolbachia can change some of the Aedes characteristic features. In practice, Wolbachia can reduce the life-span of mosquitoes by half producing a deleterous fitness effect[32]. Another feature is cytoplasmic incompatibility (CI)[22,33-35] which occurs when a Wolbachia infected male mates with an incompatible female mosquito (usually Wolbachia uninfected) producing no offspring[36]. Other features of Wolbachia which serve as liabilities in mosquitoes include: imperfect maternal transmission (IMT)[30,37] and loss of Wolbachia infection (LWI). LWI impedes the establishment of Wolbachia-infected mosquitoes and is a result of mosquito vulnerability to high temperature[38,39].

However, a novel strain of Wolbachia: wAu, has shown to produce high viral blockage whilst maintaining Wolbachia infection in Aedes mosquitoes at higher temperature[23]. Moreover, wAu allows superinfection to occur when wAu and other strains of Wolbachia co-exist in the vector host[23]. Despite these favourable features, wAu does not induce CI[23]. Although CI absence does not establish Wolbachia infected mosquitoes, the effect could be outweighed by LWI and IMT[37].

The difference in the common Wolbachia strain features are described in Table 1 below.

Table 1

Characteristics of different Wolbachia strains in Aedes mosquitoes: as defined in[22], the percentages (%) of the effects of these features are: High above 90, Medium 20 to 90, Low less than 20 and None 0, (features not detected).

Features	wAu	wMel	wMelPop	wAlbA	wAlbB
Viral blockage	High[23]	Medium[40,41]	High[29,41–43]	Medium[23]	High[44]
Maternal transmission	High[23]	High[30]	High[32,45]	High[23]	High[46]
Loss of Wolbachia infection at higher hemperature	Low[23]	High[23]	High[23]	Medium[23]	Medium[23]
Fitness cost	Medium[23]	Medium[23]	High[47,48]	High[23]	Medium[22]
Cytoplasmic incompatibility	None[23]	High[30]	High[32,45]	High[23]	High[46]

In general, the introduction of mathematical models to understand infection dynamics of diseases has long been helpful in the area of disease control[49]. A number mathematical models of Wolbachia dynamics in a mosquito population have been formulated[37,50-58]. Some of these models introduced Wolbachia strain(s) into a mosquito population and classified them into age-sturctured Wolbachia-infected and -uninfected mosquito compartments[37,53,54,57]. Ndii et al.[53], formulated a mathematical model for the Wolbachia interaction between the immature stages (aquatic stage), adult male and female mosquito populations to investigate the persistence of mosquitoes infected with Wolbachia when competing with the uninfected ones. They derived the steady state solutions and showed that parameters such as maternal transmission, reproductive, death and maturation rates drive the persistence of the Wolbachia-infected mosquito population. A similar model developed by Xue et al. considered the Wolbachia-induced fitness change and the CI effect[57]. They showed that if the basic reproduction number () of the Wolbachia-infected mosquitoes is less than one, an endemic Wolbachia infection can still occur via backward bifurcation if a sufficient number of the mosquitoes are introduced into the population. A mathematical model of Wolbachia to control dengue fever transmission[52] was developed by Hughes et al. The model showed that the use of Wolbachia has high potential to control dengue where the due to Wolbachia-infected Aedes mosquitoes is not too large in endemic areas. Another study of a Wolbachia invasive model incorporated IMT and LWI and showed that CI does not guarantee the establishment of Wolbachia-infected mosquitoes as the disadvantages derived from IMT and LWI in the production of Wolbachia-infected mosquitoes could outweigh CI[37].

Additionally, a study conducted by O’Reilly et al combining multiple modeling methods, was used to estimate the burden of dengue and map its distribution across Indonesia[59]. They predicted that there was a reduction in dengue transmission after a nationwide release of wMel-Wolbachia-infected mosquitoes. In addition, they predicted about 86% of the estimated 7.8 million annual cases of symptomatic dengue in Indonesia could be averted following a complete nationwide rollout of Wolbachia-infected mosquitoes. Recently, a modeling study presented a dengue transmission model in the presence of female wild-type and wMelPop Wolbachia-infected Aedes aegypti mosquitoes. They concluded that although the wMelPop strain reduces the lifespan of infected mosquitoes, which could be challenging to achieve replacement of wild-type mosquitoes, its optimal release ensured the replacement of wild-type mosquitoes and also reduced dengue burden in the human population[51]. A mosquito-Wolbachia model was developed by Xue et al, to compare the potential effectiveness of two Wolbachia strains (wMel and wAlbB) to control arboviral spread[60]. They observed that each of the two different strains of Wolbachia can effectively decrease the rate of arboviral transmission.

Here, we develop a general Wolbachia model capable of faithfully replicating all of the strain features described in Table 1. The general transmission model is an extention of the Wolbachia transmission model introduced in Adekunle et al.[37], which described the competitive dynamics between (wMel-like) Wolbachia-infected and uninfected mosquitoes. Despite the non-induction of CI in wAu-Wolbachia-infected mosquitoes, wAu infection is retained and able to block viral transmission efficiently compared to other strains even at high temperature. Therefore, we incorporated this feature to determine if the advantages (Wolbachia retainment) of the wAu strain outweigh the ineffectiveness of CI. This feature has not been considered in previous models. Furthermore, we incorporate imperfect maternal transmission into the model. By analysing the system via computing the basic reproduction number(s) and investigating the stability properties of the equilibrium points, the potential of the wAu strain as a viable strategy to control Aedes-borne infections can be established. The aim of this modeling approach is to support future Aedes-borne viral control programs, particularly with the introduction of new Wolbachia variants.

Methods

Model formation

Here, we investigate a modified Wolbachia transmission model studied in Adekunle et al.[37], focusing on a novel Wolbachia strain, wAu, which has high retainment, high viral blockage and does not induce CI. The mosquito population is subdivided into two groups: the uninfected mosquitoes and the Wolbachia infected mosquitoes . The term (.) can be aquatic/immature (eggs, larvae and pupae) A, male M or female F mosquitoes. In addition, we denote the aquatic/immature stages, mature male and mature female uninfected mosquitoes as , , , and Wolbachia-infected mosquitoes as , , respectively. As in Adekunle et al.[37] the model also incorporates the IMT of wAu-Wolbachia.

There are four possible mosquitoes’ mating pairs: , , and . As Wolbachia infection is maternally transmitted, and will produce uninfected offspring while and will typically produce infected offspring. However if there is imperfect maternal transmission, the two latter strategies could produce some proportions of uninfected offspring[23].

To mathematically write the system of differential equations governing the Wolbachia transmission dynamics, we express the feasible mating strategies of uninfected and Wolbachia infected mosquito populations together with their per capita egg laying rates as Eqs. (1)–(6):where , , .

Here, represents the CI effect which can be either 0 if there is no CI, or 1 if CI is present. is the effect of LWI, such that it can either be 0, if there is no Wolbachia loss or greater than zero otherwise. In Adekunle et al.[37] where CI is assumed and LWI is considered, these quantities are set to and . In our modified model, considering different strains with the exception of wAu strain, and could vary from values greater than zero onwards. However, for the wAu-Wolbachia strain, CI is ineffective and high retainment of wAu-Wolbachia infection even at high temperatures[23] is established, therefore we set and . Our model also incorporates imperfect maternal transmission generating a proportion of infected and uninfected offspring from mating of both and mosquitoes. To simplify the system, we assume that in accordance with the observed ratio of male to female mosquitoes of 1.02:1[62]. That is, we set (Fig. 1). By this, it follows that the system of ordinary differential equations (ODEs) in Eqs. (1)–(6) can be reduced to (7)–(10) which is the governing Wolbachia infection dynamics.

Figure 1

General model showing the Wolbachia infection dynamics in mosquitoes as M has been set equal to F. The green and pink compartmental polygons represent wild-type and Wolbachia-infected mosquitoes respectively. and represent the aquatic (eggs, larvae and pupae) and adult female mosquitoes for the uninfected mosquito population respectively while and represent their Wolbachia infected counterparts. The teal and red arrows illustrate the population progression of uninfected and Wolbachia-infected mosquitoes respectively. The four circles (three black and one brown) represent the mosquito mating strategies. The effect of cytoplasmic incompatibility (), i.e. for wAu and wMel strains, and respectively, is illustrated by the brown-circled . The dashed lines represent the proportion of uninfected offspring caused by imperfect maternal transmission (IMT). The blue lines depict mosquito mortality. If there is loss of Wolbachia infection (LWI), . But if there is no LWI as in wAu-Wolbachia strain, then .

To mathematically express the above schematics, we have that, the feasible mating strategies of uninfected and Wolbachia infected mosquito populations together with their per capita egg laying rates are given by the following differential system:where and . Before proceeding, we rescale each of our state variables according to the maximum total population size, which by Adekunle et al., 2019[37] is set bywhere .

The closed setwhich is a feasible region for the above system dynamics is positively invariant[37].

Hence, we let , , , , and . Also, letting , we assume a perfect maternal transmission for the reproduction outcome of mating. Therefore, the general Wolbachia model in terms of population proportions is given by Eqs. (12)–(14). Hereafter it is clear that we refer to the scaled values of each state variable and as such drop the overbar from our notation. The scaled model below now evolves in the feasible region , where .The modeling of wAu-Wolbachia transmission dynamics has not been done as this a distinction from other Wolbachia transmission models. Unlike the modeling work in Adekunle et al.[37], apart from the non-induction of CI, we considered the loss of Wolbachia infections due to seasonal fluctuation in temperature, a key dynamics that is absent in wAu strain.

Results

Analysis of the model

The above general model (11)–(14) is parametrically adjusted to simultaneously accommodate wAu and wMel Wolbachia strains. For the wAu-Wolbachia model, we set and for the wMel-Wolbachia model, we set , . The wMel-Wolbachia model parameter adjustments correspond to the model studied in Adekunle et al.[37].

Here, we want to analyse the general model(11)–(14) with arbitrary values of and to enable comparison with wAu-Wolbachia and Adekunle et al. 2019[37] models. Analysing the model for wAu, we have four steady states. The first steady state indicates non-existence of mosquitoes. The second signifies the steady state for the uninfected mosquito population only. The third describes the equilibrium point for wAu-infected mosquitoes only. Lastly, the is the equilibrium point for the co-existence of both uninfected and wAu-Wolbachia-infected mosquito populations.

Non-existence mosquito population,

The equilibrium point is trivial and is not biologically realistic. However, we can gain some insights into the competitive model dynamics by examining the case where there is no interaction between the uninfected and Wolbachia-infected mosquitoes. In other words, we want to investigate how each population would behave in the absence of the other. In particular, we derive the reproduction number of the uninfected and Wolbachia-infected mosquito populations when they do not interact:where the factor of in and stems from the choice to set M = F[62], i.e. .

These reproductive numbers determine if the uninfected and Wolbachia-infected mosquito populations will die out or persist when there is no interaction. Specifically, if and , then the two populations will die out (Fig. 2a). We observed in the decoupled case, the expressions for and are independent of the effects of CI and LWI and are therefore equivalent for both the wAu and wMel-Wolbachia strains (Fig. 2)[37].

Figure 2

Graphs showing the system trajectories in the plane for (a) wAu () and (b) wMel (, ) Wolbachia models when . The red ball point indicates the point of stability, that is representing mosquito extinction. We set and . Other parameters used for these model simulations are provided in Table 2.

Table 2

Mosquito-Wolbachia model notations.

Parameters	Description	Values (wMel)	Values (wAu)	Dimension	References
ρ_uu	Reproduction rate (egg laying rate) from mating between F_u and M_u/M_w mosquitoes	13	13	Eggs/day	[32,37,61]
ρ_ww	Reproduction rate (egg laying rate) from mating between F_w and M_u/M_w mosquitoes	10	10	Eggs/day	[30,37,61]
δ	The proportion of Wolbachia infected eggs resulting from mating between M_uF_w mosquitoes	0.95	0.95	Dimensionless	[30]
ν	The proportion of Wolbachia infected eggs resulting from mating between M_wF_w mosquitoes	1	1	Dimensionless	[37]
ϕ	The CI induction	1	0	Dimensionless	[23]
ψ	Fraction of eggs that are male	0.5	0.5	Dimensionless	[37,62]
K	Carrying capacity of the aquatic stage A	10^6	10^6	aquatic mosquitoes	[37]
σ	Loss of Wolbachia infection (LWI)	0.04	0	day^-1	Assumed
τ_u	Maturation rate of A_u aquatic stage into adulthood (per capita)	0.11	0.11	day^-1	[30,61]
τ_w	Maturation rate of A_w aquatic stage into adulthood (per capita)	0.11	0.11	day^-1	[30,61]
μ_Au	A_u Aquatic stage mortality rate (per capita)	0.02	0.02	day^-1	[57]
μ_Aw	A_w aquatic stage mortality rate (per capita)	0.02	0.02	day^-1	[57]
μ_u	F_u adult mortality rate (per capita)	0.061	0.04316	day^-1	[23,37]
μ_w	F_w adult mortality rate (per capita)	0.068	0.08079	day^-1	[23,37]

Uninfected mosquito population,

The uninfected-mosquito-only equilibrium point or Wolbachia-free equilibrium isFor to exist, we require . In addition to the uncoupled reproduction numbers ( and ) we also define the invasive reproduction number which describes the average number of secondary offspring that will become Wolbachia-infected adults after introducing a single adult Wolbachia-infected mosquito into an established Wolbachia uninfected mosquito population.

To compute , we use the next generation matrix method[63] to obtainwhere we have substituted in the definition of from Eq. (16). The invasive reproduction number is the same for both wAu and wMel-Wolbachia strains as that derived in Adekunle et al.[37]. This is because, the expression (17) clearly shows that the invasive reproductive number is not dependent on the CI effect, or LWI, .

To check if the equilibrium point is stable, we compute the Jacobian of the system and evaluate it at . In particular, letting and , yieldsTo obtain the characteristic equation of , we havewhich becomeswhereTherefore, is locally asymptotically stable if and only if and (Fig. 4). This is also consistent with the study in Adekunle et al.[37] (See Table 3).

Figure 4

This graph shows the existence and local stability regions for the equilibrium points – for the Wolbachia model (11)–(14) as a function of the and relating to the cytoplasmic incompatibility (CI), and maternal transmission (MT), i.e. magnitude of and . The yellow shaded region indicates the local stability of equilibrium. The green shaded area illustrates the local stability for the Wolbachia-free equilibrium point (). is locally stable at the red shaded part. The blue region indicates the coexistence local stability . The white region shows the existence of and and local stability of and equilibrium points. And the orange region describes the existence and local stability of and . For ; (a) describes as the boundary sits above the boundary and the arc . The co-existent equilibrium (blue), always sits in the region between these three boundaries because , and . If , then becomes stable (yellow). (b) describes similar conditions as in (a) but for . We observed that the boundary shifts up while remained stationary to accommodate more . For ; (c) describes as the relative position of boundaries in (a) flips so that boundary sits above boundary and the arc . Then, and and shows the co-existence of and (white). However, and are locally stable in the white region as and . For , and do not exist, only and do and if , and are locally stable (orange) and if , only becomes stable (red). (d) describes similar conditions as in (c) but for . It was observed that the boundary shifts up reducing the region of stability for .

Table 3

Expressions for the condition for stability associated with the equilibrium points.

Equilibrium points	Conditions for stability
	wMel[37]	wAu
(i) No mosquitoes (e_1)	R_0u<1 and R_0w<1	R_0u<1 and R_0w<1
(ii) Uninfected mosquitoes only (e_2)	R_0w|u<1 and R_0u>1	R_0w|u<1 and R_0u>1
(iii) Wolbachia-infected mosquitoes only (e_3)	R_0u|w<1 and R_0w>1	R_0u|w<1 and R_0w>1
(iv) Both mosquitoes (e_4)	δ<1, μ_u<δμ_w, R_0w>1 and R_0u>1	R_0w|u>1, R_0u|w>1, R_0w>1 and R_0u>1

Wolbachia-infected mosquito population,

The wAu-infected-only equilibrium point is . This again is consistent with Adekunle et al.[37].

For to exist we require . By computation, the invasive reproductive number with respect to uninfected mosquitoes is given as,where . Clearly, is dependent on . For the wMel-Wolbachia strain, i.e. , which is equivalent to that of Adekunle et al.[37]. However, for the wAu-Wolbachia strain, i.e. , we have a modified expression of in Eq. (18) because we do not assume CI. Therefore, for wAu-Wolbachia strain. Computing the Jacobian at , we have:The characteristic equation of is thenwhereTherefore, is locally asymptotically stable if and only if and (see Fig. 4). The condition is equivalent to that found in[37] with generalized expressions for used in place of the reduced version presented there (see Table 3).

Coexistent mosquito populations,

The equilibrium point for which both the uninfected and Wolbachia-infected populations coexist is

where and . For to exist, we require and The above conditions (i) and (ii) correspond to the cases for and respectively. Comparing these existence conditions with those found above for and , we see that condition (ii) for the existence of matches the combined existence and local asymptotic stability condition for and . In other words, , and can coexist, while always exists (see Fig. 4).

or

.

To establish whether is stable or not, we compute the Jacobian evaluated at to obtain the following characteristic equation:Let

, , , ,

, ,

then we have:In order to establish the nature of the equilibrium point , we performed numerical testing using the Monte Carlo method in[50] to verify the conditions (i) and (ii) by computing the real part of the eigenvalues of the Jacobian matrix, evaluated at . Simulation results are illustrated in Fig. 3.

Figure 3

Graphs showing the numerical testing for the stability conditions (i) and (ii) and the real part of the eigenvalues’ distribution ( and ) for : (a,b) show that always hold. (c) shows the related distribution of the real part of the eigenvalues for condition (i). (d,e) show the condition always hold while (f) shows the corresponding distribution of the real part of the eigenvalues for condition (ii).

Although the conditions (i) and (ii) indicated the existence of , Fig. 3c showed that is locally stable for condition (i) as all the eigenvalues (real part) are negative (). Whilst Fig. 3f showed that is unstable for condition (ii) as two of the eigenvalues (real part) are positive i.e. .

Numerically, we illustrated the existence and stability regions for in Fig. 4 for the two conditions (i) and (ii) relating to CI and maternal transmission (MT).

Following a modeling study of Aedes aegypti mosquitoes and normal Wolbachia (in the presence of CI only) interaction analyzed by Ferreira et al.[64], three equilibrium points: trivial (); uninfected only (); and coexistence (), were obtained. However, the Wolbachia-only equilibrium point was not computed. The established local stability conditions for and correspond to that of the wMel-like Wolbachia conditions for and respectively. For coexistent populations to persist, the reproductive number for infected mosquitoes only, must be greater than 1 and , where is the reproductive number for wild-type mosquitoes only. The model[64] also described the fitness parameter space between and , showing the change in extinction and persistence of the three equilibria when there is an increase in the initial population proportion of the Wolbachia-infected mosquitoes. Our model showed the changes in the no-mosquito, wild-type only, Wolbachia-only and coexistence population persistence and extinction in the presence and absence of CI with high and low maternal transmission (MT).

Figure 4 illustrates the existence and local stability regions for the equilibrium points , , and with respect to the reproduction numbers and as well as the relative magnitude of and . For (high MT), Fig. 4a,b describe the dynamics for (CI absent) and (CI present) respectively. Within the subset of the yellow region of these figures bounded by , and we find that only and exist. Since is unstable in this region, we expect the system trajectories to tend to the no-mosquito equilibrium . This was confirmed through numerical simulations shown in Fig. 5a. For the existence of we require , and for stability (within the blue region). But if , is stable (yellow).

Figure 5

Graphs showing the local stability for relating to the magnitude of and . The initial conditions for the state variables are , , , . We set . (a) For , where . (b) For , where . The equilibrium point is locally stable if , and .

For (low MT), Fig. 4c,d portrayed the regions of stability for and respectively. The conditions , and project the trajectiory to tend to (see Fig. 5b). In the orange region, and exist and are simultaneously locally stable as and . In addition, we have that exists where and (condition (ii)). With these conditions, exists together with and (white region). In this white region, and are locally stable even as but is unstable. Also, exists when and because the local stability of other equilibrium points is violated with these conditions. When but and , the only stable outcome is the mosquito-free (no-mosquito) equilibrium . This occurs when is less than but still close to one. In this region, uninfected mosquitoes are capable of dominating initially when introduced into a Wolbachia saturated equilibrium because imperfect maternal transmission achieves . This competitive advantage drives out the Wolbachia infected mosquitoes leaving uninfected mosquitoes only, which then are unable to sustain their population because (Fig. 5).

With the rate of high maternal transmission (MT) in the absence of CI (like-wAu), the reproductive advantage favours the production of uninfected mosquito offspring as it tends to accommodate more coexistent mosquito populations with wild-type than wMel-like strain (presence of CI) due to the presence of CI (Fig. 4a,b). Whilst, with a low MT rate, the CI presence or absence would favour Wolbachia-infected mosquitoes or uninfected mosquitoes respectively. In other words, the coexistent equilibrium point is unstable for the two mosquito populations as these conditions are equivalent to the local stabilities of both Wolbachia-free and Wolbachia-only equilibrium points (Fig. 4c,d). If , the system trajectories tend to the no mosquito equilibrium .

The conditions for the local stability of all equilibrium points are shown in Table 3 below.

Sensitivity analysis of Wolbachia model

To carry out the sensitivity analysis we investigate the model robustness due to uncertainties associated with parameter value estimations. In other words, we examine how senitive the invasive reproductive numbers are with respect to these parameters. This in turn, gives insight on influential parameters and their impact in reducing (or increasing) mosquito-type populations. To carry out this, we compute the normalized sensitivity indices of the invasive reproduction numbers with respect to the parameters used in the model.

Definition

The normalized forward sensitivity index of a variable v with respect to parameter w is defined as:Using the above formular (20), we contruct the following plots in Fig. 6.

Figure 6

Plots showing the sensitivity indices of and the model parameters.

From Fig. 6 and using the baseline parameter values for the wAu-Wolbachia strain in Table 2, it is clear that the reproductive and mortality rates for both wild-type () and wAu-Wolbachia-infected () mosquitoes and the proportion of wAu-Wolbachia-infected offspring () have the most sensitivity in the invasive reproductive numbers . Whilst for , and are the most sensitive parameters. Hence for both invasive reproductive numbers, the most sensitive parameters are and . This demonstrates that an increase (or decrease) in the mortality rate of wAu-Wolbachia-infected mosquitoes by 10% will decrease (or increase) by 10%.

Does CI outweigh the LWI ?

For most Wolbachia strains except wAu, the mating between uninfected female and Wolbachia-infected male mosquito crosses generates no viable offspring. However, Wolbachia-infected mosquitoes tend to lose their Wolbachia infection and lower their maternal transmission rate at high temperature ()[23]. With the effect of climate change gradually increasing the temperature by the day, Wolbachia strains with moderate or high temperature sensitivity such as wMel may not be able to fully maintain a sufficient frequency level to invade the mosquito population.

In our general Wolbachia mathematical model, we describe a modified version of Adekunle et al.[37]. This modification accommodates parameter adjustments for novel wAu and wMel-Wolbachia strains. For wAu, our mathematical model showed that despite the production of mosquito offspring due to CI absence, the invasive reproduction number due to infected mosquitoes remains unchanged compared to the case where CI is present, as with the wMel-like strain[37]. This further strengthened the fact that CI (inclusion or exclusion) does not guarantee Wolbachia mosquitoes’ persistence. Also, the invasive reproduction number due to uninfected mosquitoes expression for wAu is similar to wMel, except that the expression depends on CI the effect. This is because, the mosquito gender crosses due to non-induction of CI for wAu, i.e. , generates uninfected offspring with perfect maternal transmission while wMel does not. The chances of establishing Wolbachia infected mosquitoes are lower when CI is ineffective compared to when it is induced. That is, for cytoplasmic inducing wMel-Wolbachia mosquitoes, the effect of LWI outweighs CI effect as mosquitoes still lose their infections (Fig. 7). However, wAu-Wolbachia infection retainment (no LWI) in mosquitoes has shown high level of maintaining the Wolbachia frequency in the absence of CI in mosquitoes (Fig. 7). This suggests that the LWI effect outweighs CI.

Figure 7

(a) Effect of CI induction and LWI on the Wolbachia frequency level. The initial conditions for the state variables are , , , . The red line indicates Wolbachia retainment as (no CI induction) and (no LWI) which are features of wAu-Wolbachia strain. The blue and black dashed lines (for wMel-Wolbachia strain) illustrate CI induction and LWI i.e for and respectively. Parameters for were used in these simulations. (b) Shows the dominance of wAu-Wolbachia infected to uninfected adult mosquitoes due to the retainment of Wolbachia infections (not affected by seasonal varying LWI). The wAu-Wolbachia-infected mosquitoes dominates when there is no CI and LWI (red line). (c) For wMel-Wolbachia-infected mosquitoes, the effect of seasonal varying loss of Wolbachia infection is shown as infections rise and drop continuously due to LWI and CI induction .

The LWI rate which is dependent on the seasons of the year can be modeled by a sinusoidal equation:where is the maximum value of the seasonal variation in LWI, and is the phase shift which aligns the model with the seasonal change.

The effects of CI () and LWI () as features of wAu and wMel Wolbachia strains are shown in Fig. 7. For the total mosquito population, wAu-infected mosquitoes () reach the maximum frequency after approximately 250 days. To see the effect of CI induction and slight LWI i.e. , , for and , the Wolbachia frequency level oscillates between (0.8 and 1) and (0.6 and 1) respectively. That is, there is a 20% and 40% drop in the frequency level of Wolbachia when is at and respectively. This showed that, despite CI induction, LWI reduced the contribution of CI to the Wolbachia invasion (Fig. 7a). Therefore, the LWI gains highly outweigh the CI effect. By this, our analysis suggests that an increase in LWI in the presence of CI results in a drastic decrease in the Wolbachia frequency level (Fig. 7a). On the other hand, Fig. 7b showed the effect of LWI and CI with respect to the competitiveness between and . We observed that the population dominates the when there was no CI induction and Wolbachia infection is retained, that is, (Fig. 7b). However, if CI induction occurs with loss of Wolbachia infections, then the seasonal varying effect occurs as seen in Fig. 7c.

Discussion

In this work, we modelled and investigated a general Wolbachia model that contained the transmission dynamics of wAu and wMel Wolbachia strains in Aedes mosquitoes as special cases. These transmission dynamics described the competition between the novel wAu-Wolbachia infected Aedes mosquitoes and wild-type mosquitoes and compared the dynamics with the invasive properties of the popular wMel-Wolbachia infected mosquitoes. We first derived the Wolbachia infection-status reproduction numbers for our wAu-Wolbachia model and used them to establish the conditions for the local stability of the equilibrium points for the wAu-Wolbachia invasive model. The reproduction number associated with the uninfected mosquitoes shows the reproductive advantage that the wild type has over the wAu strain. The comparison of the wAu-Wolbachia model (CI and LWI absent) and wMel-Wolbachia model (CI and LWI present) showed that the wAu strain has the potential of compensating for the undesirable features of the wMel strain.

Additionally, this study has reviewed the main features of different Wolbachia strains (Table 1) and shown that the wAu Wolbachia strain is a promising candidate for efficient Aedes-borne arboviral transmission control. Moreover, we analyzed the system dynamics of a general Wolbachia invasion model and determined the regions of local stability for each of the identified equilibrium points, highlighting the regions in parameter for which Wolbachia-infected mosquito populations persist or go extinct. This work modelled the general Wolbachia dynamics which can accommodate various Wolbachia characteristics regarding the presence or absence of CI and seasonal changes, unlike Adekunle et al.[37], which considers only the presence of CI. We also investigated the advantages gained from CI and LWI. This study has demonstrated that despite the absence of CI, the Wolbachia frequency level will drop as much as tenfold of the percentage of Wolbachia infection lost. We showed that the advantage of Wolbachia retainment in mosquitoes strongly outweighed the negative impact of CI indicating wAu Wolbachia strains may be suitable for arboviral control. Therefore, this modeling work contributes to the previous studies[37,54,57,64,65] and helps close the gap between ways of maintaining the Wolbachia frequency levels in the absence of LWI and CI.

One implementation question for using the wAu strain as a replacement of the wMel strain is whether the wAu strain is self-sustaining, given that it does not induce CI. In this work, the equilibrium points for the wAu-Wolbachia model are the same as that for the wMel-Wolbachia model except that stricter conditions are required to satisfy the wAu-Wolbachia model equilibrium points. These more stringent conditions translate to additional resources such as the continuous introduction of a larger scale of wAu-infected mosquitoes to ensure replacement[66]. Thus, the wAu strain is a promising alternative strain as it does not suffer from LWI due to high weather temperature and is highly effective in preventing the transmission of the arbovirus[23,39,67]. Otherwise, combining the two strains may also be a good strategy.

There are limitations associated with any mathematical modeling work, and this study is not exempted. We first assumed the same mosquito gender ratio and expected this proportion to be constant over time. This assumption may be true in a laboratory setting[62], but not necessarily true in a natural mosquito habitat. However, similar conclusions are expected to be reached as the Wolbachia model reduction accurately reproduces the dynamics of the full system[68]. Secondly, we assumed that the absence of CI implies that cross mating resulted in offspring that are uninfected. This may not be true as a small proportion of the offspring may be Wolbachia infected[23]. If that is the case, then it means that lesser resources will be required to use the wAu strain as a Wolbachia-based control strategy. Lastly, we assumed the seasonality affects the associated parameters for the wMel dynamics. However, for the wAu strain, it is not affected by seasonality as wAu-Wolbachia infections are retained at high temperature.

Although several studies[22,38,42,57] have demonstrated that CI drives the persistence of Wolbachia-infected Aedes mosquitoes, these studies neglected the impact of Wolbachia loss in mosquitoes. The CI drive has been shown in four mating lines (see Fig. 1) involving a Wolbachia-transinfected Aedes mosqiutoes mating with wild-type mosquitoes. One of the mating lines for which Wolbachia-infected male and uninfected female mosquitoes produced no viable offspring (via CI) truncates the uninfected offspring from being produced as infection is maternally transmitted. With the exception of the mating between the uninfected male and female mosquito line, all other mating lines produce Wolbachia-infected offspring leading to persistence. In addition, high temperature affects these Wolbachia-infected mosquitoes as they lose their infection due to the unfavourable weather conditions. However, mosquitoes infected with the wAu-Wolbachia strain have been shown to not only block arboviral transmission efficiently, but also retain the Wolbachia infection at typically unfavourable high temperatures. This retainment of infection in mosquitoes strongly outweighed the absence of CI for the wAu strain in the establishment and dominance of wAu-Wolbachia infected mosquitoes.

While vaccine implementation may have been highly effective on dengue seropositive persons in high transmission areas[11,12], the introduction of Wolbachia-infected mosquitoes in low and moderate arboviral endemic areas has also effectively shown successful reduction in dengue burden[43,51,59,69]. Given that these two strategies could reduce the transmission of Aedes-borne diseases, in particular, dengue depending on the transmission level, a modeling study by Ndii[70] proposed the use of these combined strategies and compared their effectiveness. The author showed that, Wolbachia performs better in the presence of low vaccine efficacy, but is outperformed otherwise[70]. Therefore combining the two strategies may be useful, however understanding both the temperature and seasonality effects on Wolbachia intervention programs, and serotypic differences relating to cross-protective immunity to investigate vaccine efficacy is necessary for the reduction and control of Aedes-borne arboviral disease transmission.

In conclusion, we have shown that the wAu-Wolbachia strain could be effective in controlling arbovirus transmission, as its advantages in terms of Wolbachia infection retention in mosquitoes may outweigh the absence of CI. This could prove even more promising, especially as the temperature increases due to climate change. Although wMel and wAlbB-Wolbachia strains only have been rolled out in natural mosquito habitats in replacement programs, combining these strains with wAu is worth exploring.