Modeling the Peste des Petits Ruminants (PPR) disease transmission dynamics with impacts of vaccination and restocking in small ruminant population in Amhara region, Ethiopia Authors: Yibekal Walle, Joseph YT Mugisha, Dawit Melese, Haileyesus Tessema Published: 2024 ABSTRACT AbstractPeste des Petits Ruminants (PPR) is a highly contagious transboundary viral disease of small ruminants with significant economic implications caused by the Peste des Petits Ruminants virus. This study employs mathematical modeling to investigate the impact of imperfect PPR vaccines and restocked small ruminants on the transmission dynamics of PPR. A deterministic mathematical model is developed by incorporating vaccinated and restocked subpopulations into the classical SEIR model. The influence of infected animals introduced through restocking on vaccination efficacy in preventing PPR spread is examined. The global dynamics of equilibrium points in the model are analyzed using the Lyapunov-LaSalle invariance principle. Parameter values for numerical simulations are estimated based on monthly PPR data from the Amhara regional state in Ethiopia, obtained from the Ministry of Agriculture. The basic reproduction number (Rv) is calculated to assess the level of PPR in the small ruminant population, and sensitivity analysis of parameters is performed on Rv. The analytical and numerical results reveal that infected restocked small ruminants significantly facilitate the spread of PPR in the population. Moreover, even with high efficacy vaccination, the system exhibits a unique asymptotically stable endemic equilibrium. These findings emphasize that appropriate vaccination alone is insufficient to control and eradicate PPR in the region. Implementing strict movement restrictions and biosecurity measures are necessary. These findings provide valuable insights for national policymakers in achieving the regional and national targets for PPR eradication by 2027.Keywords: Global stability, PPR disease, Vaccination efficacy, Restocking Introduction Peste des petits ruminants (PPR) is a highly contagious non-zoonotic infection caused by the Peste des Petits Ruminants virus, which belongs to the genus Morbillivirus in the family Paramyxoviridae [1]. Primarily affecting goats and sheep, PPR is closely related to the bovine and buffalo rinderpest virus, as well as the human measles virus [2], [3]. Clinical symptoms of PPR include high fever (pyrexia), conjunctivitis, oculo-nasal discharges, necrotizing and erosive stomatitis, diarrhea, and bronchopneumonia, which can ultimately lead to the death of the animal or recovery from the disease [4]. With a high transmission rate, PPR can have a devastating impact on small ruminant populations, often resulting in morbidity and mortality rates close to 100% in naive populations [3]. Currently safe and effective live attenuated vaccine is available against the PPR virus. The PPR vaccine is administered to susceptible small ruminants, such as sheep and goats, through injection or intranasal administration. Vaccination aims to induce active immunity in animals, stimulating the production of antibodies that can neutralize the virus and protect against future infections [5], [2], [6]. The spread of Peste des petits ruminants occurs through close contact between infected animals, with the virus shed in secretions (such as nasal and salivary) and excretions (such as feces) of infected animals. According to the Food and Agriculture Organization (FAO), PPR affects approximately 30 million animals across 70 countries worldwide, with 60% of these countries situated in Africa and the remaining 40% in the Middle East and Asia [7]. First described in 1942 in Côte d'Ivoire, West Africa, PPR has now become widespread throughout Asia and most of Africa, posing a major constraint on small ruminant production and welfare in these regions. This disease presents a risk to the food security and livelihoods of the world's poorest people, as sheep and goats are often valuable resources for them [3]. In Ethiopia, PPR was first clinically suspected in 1977 in a goat herd in the Afar region, and subsequent evidence of its presence was reported in 1984 and confirmed in 1991 [8], [9], [6]. Despite efforts by organizations like the FAO and the World Organisation for Animal Health (OIE) to control PPR in several countries, including Ethiopia, the disease continues to spread throughout Africa and Asia, burdening livestock keepers and veterinary resources in affected areas. In the Amhara regional state of Ethiopia, retrospective data analysis from 2010 to 2014 revealed a high occurrence of PPR, with a total of 63 outbreaks, 3,978 cases, and 919 deaths reported [9]. This data highlights the significance and widespread nature of PPR in the Amhara region [9], [5]. Several risk factors contribute to the spread of the PPR virus into disease-free regions. These factors include herd size, mixed herds of susceptible animals (goats and sheep), the introduction of animals of unknown origin without proper health guarantees or quarantine periods, the return of unsold animals from markets, religious festivals leading to increased trade movements and the establishment of slaughter centers, as well as the trading, loaning, and gifting of animals [3], [7], [10]. These activities, collectively known as restocking, can inadvertently introduce infected animals into previously disease-free areas, resulting in the rapid spread of PPR and subsequent outbreaks [9], [11]. Given the transboundary nature of PPR [12], it is crucial to incorporate restocking cases into the analysis to understand their impact on PPR transmission dynamics, develop effective prevention mechanisms, and assess whether restocking of small ruminants will affect the implementation of PPR vaccines. Furthermore, it is important to evaluate the current control measures in place and their effectiveness in controlling or eradicating the disease. In the existing literature, several statistical studies on Peste des Petits Ruminants (PPR) have been conducted. These studies primarily focus on epidemiological aspects, spatial distribution, and risk factors, employing statistical tools such as logistic regression models and univariate or multivariate analyses. Notably, the authors of the study [2] explored the spatial distribution of PPR in Ethiopia and examined the risk factors associated with positive serological status. The results indicated that the circulation of PPR Virus (PPRV) is highly heterogeneous. In their national serological survey, a significant sample size was collected in the Amhara region, comprising 43.9% of the total samples. Research presented in [6] provided insights into the prevalence and distribution of PPR in the Amhara region through cross-sectional and retrospective studies aimed at estimating seroprevalence, identifying risk factors (including species, sex, agro-ecology, and study location), and assessing the distribution of PPR. Their findings revealed that PPR is endemic in the area, with PPR seropositivity and disease outbreaks correlating with identified risk factors. The study underscored the necessity for continuous surveillance and vaccination efforts to mitigate the disease's impact. A similar study [13] was conducted in Afar region (adjacent of Amhara region), focusing on the seroprevalence and associated risk factors of PPR in selected districts. The results indicated a higher seroprevalence of PPR virus infection, confirming that PPR is endemic in that region. These studies highlight the prevalence and circulation of the disease in the respective areas. However, to evaluate the effectiveness of control measures in relation to other risk factors associated with the restocking of small ruminants, the present study adopts a mathematical epidemiology modeling approach based on the findings of these statistical investigations and prevalence data of the disease that reported the last nineteen years in the region. Mathematical epidemiological modeling is a widely used tool for simulating, predicting disease behavior, evaluating control, and creating potential scenarios to estimate the effects of contagious disease outbreaks, such as Peste des Petits Ruminants (PPR), Anthrax [14], Brucellosis [15], and Foot-and-Mouth Disease (FMD) [16], in populations of domesticated animals. Policymakers can leverage the data generated from these scenarios to implement disease control strategies and plan for early and ongoing efforts in disease management and eradication [7]. However, there have been comparatively few mathematical epidemiological studies [3], [7], [17] focused specifically on modeling PPR, with respect to other infectious animal diseases. Notably, the study by Fournie et al. [3] demonstrated the potential for disease transmission and the possible coverage of vaccination to eradicate PPR in two agro-ecological zones (highland and lowland areas) of Ethiopia, employing a meta population model that integrates both stochastic and deterministic approaches within the SIR framework. To our knowledge, no mathematical or statistical study has specifically examined how the restocking of small ruminants affects the dynamics of PPR and the interplay between restocking and vaccination. This article aims to model the transmission dynamics of PPR, considering the impact of restocking on small ruminant populations, and evaluate the effectiveness of vaccination in suppressing the spread of PPR in the Amhara region of Ethiopia. We incorporate a restocked class into a classical SVEIR deterministic compartmental model of PPR to investigate the impact of restocking on disease transmission dynamics. The model takes into account factors such as the introduction of animals of unknown origin and trade movements. By simulating different scenarios, policymakers can assess the effectiveness of control measures and vaccination strategies in managing and eradicating PPR. Model formulation In this section, a nonlinear epidemic model with restocking effect is formulated. The population of small ruminants with homogeneity N(t) is divided into susceptible (S(t)), vaccinated (V(t)), exposed (E(t)), infectious (I(t)), restocked (B(t)) and recovered (R(t)) compartments. Therefore the total population becomes The population of susceptible ruminants is generated through the birth of newborn animals (at the rate Λ). The susceptible population acquires PPR infection through effective contact with PPR-infected ruminants at a mass action incidence rate of βSI, where β represents the disease transmission rate relative to infected ruminants in the I(t) class. This population is reduced by vaccination at a rate of ω and by natural death at a rate of μ. Additionally, the population is further increased through restocking from the market and other sources, with a fraction α1 of the restocked population B. Therefore, the dynamics of this class can be described by the following equation: To describe the vaccination class, we assume that a fraction of susceptible small ruminants receive vaccinations at a rate of ω and subsequently join the vaccinated class. In cases where vaccination delivery is inadequate, small ruminants may not only receive vaccinations but also lose their immunity. However, there is a probability that this immunization will reduce the likelihood of a successful infection by ϵ. If ϵ=0, small ruminants are fully protected from PPR by the vaccine, whereas if ϵ=1, they are not protected at all. The natural death rate (μ) lowers this population. A fraction of newborn animals denoted by ρΛ (where 0≤ρ≤1) are successfully vaccinated and move to the vaccinated class. The population in the exposed compartment is generated when susceptible ruminants become infected after interacting with PPRV-infected ruminants. The population is decreased by the transfer rate σ to the infected class and by the natural death rate μ. Additionally, a fraction α2 of the restocked population B transitions into the latent compartment. Thus, we can describe the dynamics of this class using the following equation. The population of infected ruminants is reduced by the natural death rate μ, the rate of death induced by the disease δ, and the restocking process, where a fraction α2 of the population B is added in unit time. Additionally, individuals from this class are transferred to the recovered class at a rate of γ. Therefore, the dynamics of this class can be described by the following equation. The restocked class B is continuously replenished through various sources such as returns from markets, trading, loaning, giving animals, restocking programs, and other unknown origins. This recruitment process is represented by the recruitment rate κ, ensuring a constant influx of newly restocked animals. However, since the restocked animals come from different unknown origins without undergoing health guarantee or quarantine, there is a possibility of introducing infected individuals into the population. Hence, the restocked class B may contain infected animals. On the other hand, the outflow from the restocked class is determined by the rates α1, α2, and α3, which represent the proportions of animals moving from the restocked class to the susceptible class S, the exposed class E, and the infected class I, respectively. Additionally, there is a natural death rate denoted as μ contributing to the outflow from the restocked class. Taking these factors into account, we can describe the rate of change of the restocked class B with respect to time using the following equation: Finally, the recovered class is generated by transfer rates γ and is reduced by the natural death rate μ. Thus, we obtain the following equation for this class. All parameters in the model (7) maintain non-negative, and their biological meanings are described in Table 1 so that the total host population remains bounded. Table 1Parameter description of model (7).Table 1ParameterDescriptionΛSmall ruminants per capita birth rateκA constant flow rate of small ruminant into Restocked populationβEffective contact rate (PPR transmission rate)μNatural mortality rateωThe rate at which susceptible individuals are vaccinated1 − ϵVaccine efficancyδDisease-induced mortality rateσThe rate at which the exposed population becomes infectiveγRates of natural recovery from infected classρFraction of recruited small ruminants that are vaccinatedα1The rate of flow from restocked class to susceptible classα2The rate of flow from restocked class to exposed classα3The rate of flow from restocked class to infected class Parameter description of model (7). Based on the above assumptions of the model, Fig. 1 depicts the transfer diagram between compartments. The compartmental diagram for PPR disease transmission model (7). Combining all the differential equations formulated in (1)-(6) gives the following system of equations where, we assume that 0≤α1+α2+α3<1. Model analysis Well-posedness of the model In this section, it is essential to provide evidence of the existence of a solution that is not negative, as well as to establish the uniqueness and boundedness of the solution for the mathematical and epidemiological interpretation of the model (7). The total population N(t) at any time t, can be determined by adding all equations in (7), we obtain where Θ=Λ+κ and solving the differential inequality (8) with initial condition N(0)=N0 using ‘Groonwall's Inequality’ gives then As a result, the total population N(t) experiences fluctuations over time. When there is no disease or δ=0 (indicating the absence of disease-dependent mortality), the total population size N(t) converges to Θμ. Since the spread of the disease within the population reduces N(t), it can be inferred that N(t) is within the range of [0,Θμ]. Therefore, the feasible region of the model (7) is defined as follows: For any initial conditions S0,V0,E0,I0,B0,R0≥0 there exists a unique solution S,V,E,I,R respectively, defined on [0,∞) , to (7) satisfying and By applying the Theorem from the Appendix of [18], we can express system (7) in the vector form as follows: where x=(S,V,E,I,B,R) and f(x)=(f1(x),f2(x),f3(x),f4(x),f5(x),f6(x)). It has been explicitly verified that if x∈R+6 and xi=0, then for every i=1,...,6, f(xi)≥0. For instance, considering the fifth and the first equation in (7), we have f5(x)=dBdt=κ>0 when B=0, and f1(x)=dSdt=(1−ρ)Λ+α1B≥0 when S=0. Let x be a solution with values in R+6, defined in the interval [0,∞). To show that x is bounded on [0,∞), it suffices to observe that N(t)≤Θμ for all 0≤t<∞. Since S,V,E,I,B, and R are nonnegative and sum up to N, they are bounded by Θμ. The claim is supported by Theorem A.1 in [18]. □ Every solution of model (7) with positive initial data remains non-negative for all t>0 . Furthermore, the feasible region of the model is defined as follows: is a positively invariant region with respect to (7) . Let t1=sup{t>0:S,V,E,I,B,R>0,[0,t]}, so t1>0. (we follow the same procedure as [19].) From the fifth equation of model (7), we have Integrating the inequality (9) from 0 to t1, and solving for B(t) at t1, yields Considering the first equation in (7), Integrating the inequality (10) from 0 to t1, and solving for S(t) at t1, yields Therefore, B(t),S(t)>0 for all t>0. By similar augments above, it can be verified that V(t)>0,E(t)>0,I(t)>0,R(t)>0 for all t>0. And, from Lemma 3.1 the boundedness of the solution for (7) and N(t)≤Θμ suggests that the feasible region can ensure that each solution (S(t),V(t),E(t),I(t),B(t),R(t)) of the model (7) with positive initial data in Γ remains unique, positive and bounded for all t>0 and is attracted in Γ. That is, Γ is a positively invariant region. □ Existence and stability analysis of equilibria For the case of no infective influx of restocked small ruminants Considering that there is no influx of infective restocked small ruminants, i.e., α2=α3=0, and given that the state variable R does not impact the remaining equations, the system (7) can be reduced to the following five-dimensional system. and It is apparent that model (11) permits the existence of another biologically feasible region, which is also positive-invariant and attracting such that model (11) is well-posed in Γ1. Similar to the original model, but with a few minor modifications, the aforementioned result can be verified. The PPR-disease free equilibrium state ξ0, of the system (11) can be obtained as: The local stability of ξ0 can be determined by calculating the control reproduction number Rv using the method of the next-generation operator developed by Van den Driessche and Walmough [20]. The control reproduction number Rv serves as a threshold parameter that predicts the spread of the disease within the population. It quantifies the expected number of new cases of PPR generated by a single infected small ruminant in a completely susceptible population within a region with a vaccination program. If Rv<1, the infection will die out. Conversely, if Rv>1, the disease will persist in the population. The control reproduction number of the system (11) is given as: A similar threshold quantity known as the basic reproduction number R0, is obtained by setting ω=0 and ρ=0 in Rv and given by The quantity R0 [20] measures the expected number of secondary cases generated by an index case in an unvaccinated, equilibrium population (Θμ,0,0,0,κα1+μ). With this definition for R0, Rv can be expressed as Furthermore, setting Rv=1 and solving for ω gives the threshold vaccination rate, ωc: (i).By considering that ω≥0 and 0≤ϵ≤1, we can establish the relationship R0≥Rv≥ϵR0. This implies that if ϵR0>1 (which is equivalent to ϵ>1R0=ϵc), then Rv>1. Therefore, even with vaccination efforts, it is not possible to achieve Rv<1. The critical value for the reduction rate of infection due to vaccination is defined as ϵc. By considering that ω≥0 and 0≤ϵ≤1, we can establish the relationship R0≥Rv≥ϵR0. This implies that if ϵR0>1 (which is equivalent to ϵ>1R0=ϵc), then Rv>1. Therefore, even with vaccination efforts, it is not possible to achieve Rv<1. The critical value for the reduction rate of infection due to vaccination is defined as ϵc. (ii).Note that Rv≤R0 holds true, with equality occurring only when ω=0 and ρ=0. It is observed that in the case of imperfect vaccines, where ω>0 and 0<ρ≤1, the effect is a reduction in the reproduction number of the disease. Note that Rv≤R0 holds true, with equality occurring only when ω=0 and ρ=0. It is observed that in the case of imperfect vaccines, where ω>0 and 0<ρ≤1, the effect is a reduction in the reproduction number of the disease. The PPR disease-free equilibriumξ0is locally asymptotic stable ifRv<1but unstable forRv>1. To prove this theorem, we employ the linearization method. The Jacobian matrix of system (11) evaluated at the PPR-disease free equilibrium state ξ0 is as follows: We obtain the characteristic polynomial equation of the Jacobian matrix, which is given by where Since all parameters in (13) are positive, some roots of the characteristic polynomial in (13) are λ1=−μ, λ2=−(ω+μ), and λ2=−(α1+μ), all of which have negative real parts. It should be noted that b1>0 and b0>0 only if Rv<1, as determined by the Routh-Hurwitz criterion in [21]. According to this criterion, the other two roots of (13), λ4 and λ5, will also have negative real parts if both b0 and b1 in (14) are positive. Consequently, the PPR disease-free equilibrium state ξ0 will be locally asymptotically stable if Rv<1. □ If Rv≤1 , ξ0 is globally asymptotically stable in Γ1 . Following the same idea in [19], we take the candidate Lyapunov function L(t)=E(t)+(σ+μσ)I(t). Now, calculate the time derivative of L(t) along the solution of model (11), we find If Rv≤1, then L′(t)≤0 for (S,V,E,I,B)∈Γ1 sufficiently close to (S0,V0,0,0,B0) and L′(t)=0 holds only when I=0. Since {ξ0} is the largest invariant set in the subset of Γ1 where L′(t)=0. Its globally asymptotically stability follows from local stability of ξ0 and Lasalle's invariance principle [22]. □ The endemic equilibrium of model (11) can be given: and from equations (15) and model (11), I⁎ is the positive root of the cubic polynomial: where According to Equation (16), it can be deduced that a>0 and c<0 if Rv>1. Consequently, the quadratic polynomial H(I⁎) possesses two roots with opposite signs. Therefore, it can be concluded that the model (11) has a unique positive endemic equilibrium ξ⁎=(S⁎,V⁎,E⁎,I⁎,B⁎) when Rv>1. Model(11)has a unique endemic equilibriumξ⁎if and only ifRv>1. The endemic equilibrium ξ⁎ is locally asymptotically stable if Rv>1 . In addition, the model (11) exhibits the forward bifurcation. In line with Theorem 4.1 of [23], we employ center manifold theory [24]. Now, let us express the system (11) in vector notation: where x=(x1,x2,x3,x4,x5)=(S,V,E,I,B) and f=(f1,f2,f3,f4,f5). Let β be the bifurcation parameter, then system (11) becomes with Rv=1 corresponding to β=β⁎=μ(ω+μ)(σ+μ)(δ+γ+μ)(α1+μ)σ(ϵρΛ(ω+μ)(α1+μ)+(ϵω+μ)(Λ(1−ρ)(α1+μ)+α1κ)). The Jacobian matrix of the system (11) at the PPR disease-free equilibrium point ξ0 when β=β⁎ is given by where ξ0=(x10,x20,x30,x40,x50) corresponding to (S0,V0,E0,I0,B0). We obtain the characteristic polynomial equation of the Jacobian matrix (18), which is given by It is clearly shown in (19) that 0 is a simple eigenvalue of J(ξ0)|β=β⁎, and the other eigenvalues are real and negative (ξ0 is a non-hyperbolic equilibrium). Hence, Theorem 4.1 in [24] can be used to analyse the dynamics of (11) near β=β⁎. In particular, it will be used to show locally asymptotic stability of the endemic equilibrium of (11) for β near β⁎. It can be shown that a right eigenvector associated with zero eigenvalue is u=(u1,u2,u3,u4,u5)T, where We can see in (20) that u1<0;u2<0;u3>0; and u4>0. The left eigenvector v=(v1,v2,v3,v4,v5) of J(ξ0)|β=β⁎ (associated with the zero eigenvalue) can be obtained as Here we have taken into account the expression for β⁎, where u4 and v3 are calculated to ensure that the eigenvectors satisfy the condition u.v=1. Based of this condition we have u4v3=σδ+γ+2μ. Next, we will compute the values of a and b based on Theorem 4.1 of [24]. From equation (21), since v1=v2=v5=0, we only need to compute the partial derivatives of the f3 and f4 in (17) at the PPR disease-free equilibrium. For system (17), the associated non-zero partial derivative of f3 and f4 at the PPR disease-free equilibrium is given by with where all partial derivatives of f3 with respect to other variables and all partial derivatives of f4 are zero. Then we have Thus, the signs of a and b in (22) and according to item (iv) of Theorem 4.1 in [24], we can conclude that ξ⁎ is locally asymptotically stable if Rv>1. Furthermore, when the control reproduction number is equal to one, a change of stability occurs. Specifically, the PPR disease-free equilibrium becomes unstable, and a non-trivial (endemic) equilibrium arises and remains stable. Therefore, system (11) with parameter values outlined in Table 2 exhibits a forward bifurcation, as depicted in Fig. 2, when Rv=1. The presence of a forward bifurcation in our model indicates that the basic reproduction number becomes the sole determinant for the eradication of PPR disease. As long as we can maintain a condition of Rv<1, PPR disease will always vanish from the small ruminant population. □ Table 2Parameter values from estimation, literature sources, and assumptions for the case of zero infective restocked animals.Table 2ParameterValue (unit of per month) with CISourceβ2.0914e-7 (CI:1.8985e-7 2.2843e-7)Estimatedσ0.6058 (CI:0.5747 1.7864)Estimatedγ0.4753 (CI:0.0615 1.5664)Estimatedμ0.0527[30]ω0.5694 (CI:0.3328 1.8072)Estimatedδ0.3[29]κ224674[30]α10.01AssumptionΛ645990Estimatedρ0.1681Assumptionϵ0.3[3] Parameter values from estimation, literature sources, and assumptions for the case of zero infective restocked animals. The forward bifurcation diagram for model (11). For the case of a non-zero influx of infective restocked small ruminants In this section, we explore the dynamics of the model (7) under the scenario of a non-zero influx of infected restocked small ruminants within the feasible region Γ. As (dEdt+dIdt)E=I=0>0, it follows that model (7) lacks a disease-free equilibrium of PPR. Consequently, the system may have a unique endemic equilibrium. We analyze the global dynamics of the disease based on the endemic equilibrium point using an approach similar to that employed in prior studies [25], [26]. Let ξ¯=(S¯,V¯,E¯,I¯,B¯) represent the endemic equilibrium of system (7) and satisfy the following equilibrium equations where α¯=α1+α2+α3+μ. The expression for the endemic equilibrium can be given by: and from equations (24) and (23), I¯ is the positive root of the cubic polynomial: where Notice that when α2=α3=0, or α=0 in (25), the equation G(I¯,α1,0)=0 reduces to I¯F(I¯)=0. Consequently, we have I¯=0 for the disease-free equilibrium (DFE), and F(I¯)=0, which is equivalent to equation (16). This equation has a unique positive root for the endemic equilibrium if c0¯<0, which is equivalent to Rv>1. Furthermore, we observe that G(I¯,α1,α)0 and I¯>0. To analyze the number of positive roots for G(I¯,α1,α)=0, we need to consider three cases based on the signs of b¯ and c0¯ in (26). A similar approach has been employed in the study of the model [27], [28]. In the case where c0¯<0, according to Descartes' Rule of Signs, the equation G(I¯,α1,α)=0 has at most one positive root, regardless of the sign of b¯. This result holds because a¯>0, c¯<0, and d¯<0. On the other hand, equation F(I¯)=0 has a unique positive root denoted as Iˆ. Consequently, we can deduce that G(Iˆ,α1,α)<0, as G(I¯,α1,α)0. Furthermore, it can be observed that G(I¯,α1,α)→∞ as I¯→∞, which confirms the existence of a positive root for G(I¯,α1,α). In the case where c0¯>0 and b¯>0, let r1, r2, and r3 denote the roots of G(I¯,α1,α). In this scenario, given that a¯>0 and b¯>0, it follows that r1+r2+r3=−b¯/a¯<0 and r1r2r3>0. Consequently, it can be concluded that G(I¯,α1,α) has a unique positive root for any α>0. For the final case where c0¯>0 and b¯<0, if b¯2−4a¯c0¯>0, Descartes' Rule of Signs indicates that F(I¯) possesses two positive roots. Thus, if b¯2−4a¯c0¯>0, G(I¯,α1,0) has two positive roots. Since G(I¯,α1,α) is a cubic function of I¯ and a decreasing function of α for I¯>0, it can be observed that there exists a positive value αˆ such that G(I¯,α1,α) has three positive roots if 0<α<αˆ, two positive roots (one with multiplicity 2) if α=αˆ, and a unique positive root if α>αˆ. It should be noted that αˆ<α0, where To illustrate this, suppose α≥α0. If G(I¯,α1,α) has exactly one real root, then from r1r2r3=−d¯a¯>0, it can be deduced that this root must be positive. If the roots of G(I¯,α1,α) are all real, considering that r1r2+r1r3+r2r3=c0¯−αB¯β(ϵ(ω+μ)+μ)(σ+μ)(δ+γ+μ)a¯≤0 when α≥α0, it can be observed that G(I¯,α1,α) also possesses a unique positive root in this case. These findings are summarized below. i.Ifω<ωcorα>α0, orc0¯<0, then the model(7)has a unique positive endemic equilibrium. Ifω<ωcorα>α0, orc0¯<0, then the model(7)has a unique positive endemic equilibrium. ii.Ifω>ωcor,c0¯>0,b¯>0then the model(7)has a unique positive endemic equilibrium forα>0. Ifω>ωcor,c0¯>0,b¯>0then the model(7)has a unique positive endemic equilibrium forα>0. iii.Ifω>ωc,b¯<0, andb¯2−4a¯c0¯>0, then there is a positiveαˆ<α0such thatG(I¯,α1,α)has a unique positive root ifα>αˆ. Ifω>ωc,b¯<0, andb¯2−4a¯c0¯>0, then there is a positiveαˆ<α0such thatG(I¯,α1,α)has a unique positive root ifα>αˆ. In general, since G(I¯,α1,α) is a decreasing function of α, limI¯→0+⁡G(I¯,α1,0)=0, and limI¯→∞⁡G(I¯,α1,α)=∞, it follows that G(I¯,α1,α) has at least one positive root, and hence at least one endemic equilibrium always exists. For system(7)the endemic equilibrium(S¯,V¯,E¯,I¯,B¯)is globally asymptotically stable on the set Γ. Let ξ¯=(S¯,V¯,E¯,I¯,B¯) denote the endemic equilibrium. Consider a Lyapunov function where g(x)=x−1−ln⁡x and D=σ+μσ,α¯=α1+α2+α3+μ. Note that L(x) is positive definite with respect to ξ¯ because x¯g(xx¯) achieves a global minimum value of 0 at x=x¯. The derivative of L along a solution (S(t),V(t),E(t),I(t),B(t)) of the system (7) is expressed as follows: To establish the characteristics of a Lyapunov function for which dLdt≤0, we consider each terms in (27) independently then sum them up. By the first equations in (7) and (23), it follows that By the second equations in (7) and (23), it follows that Similarly, by (7) and endemic equilibrium equations in (23), we have Since D=σ+μσ, we have used relation DσE¯=α2B¯+βS¯I¯+ϵβV¯I¯. From the fifth equation in (7), we have Substituting (28)- (32) into (27) and rearranging terms, we obtain From (23), we have the relations: Thus, the inequality in (33) becomes: By the inequality of arithmetic and geometric means: and since all the model parameters are non-negative, it follows that L′≤0. As a result, L′=0 holds only at ξ⁎⁎, and L′ is negative definite with respect to ξ¯. Standard Lyapunov stability theorem implies that ξ¯ is locally asymptotically stable and globally attracting in Γ. The global stability of ξ¯ also implies that ξ¯ is unique in Γ. □ Parameter estimation In this study, we estimate the parameters in the system (7) using monthly PPR disease incidence data that span from January 2019 to August 2022 in the Amhara regional state, Ethiopia. Data were acquired through a filtering process that specifically targeted the incidence of PPR within the Amhara region, using the Ethiopian National Animal Disease Surveillance System (NADSS). In particular, the data were sourced from the monthly disease outbreak and vaccination activity reporting (DOVAR), which operates under the authorization of the Minister of Agriculture. Furthermore, we incorporated data on the death rate attributable to PPR from the World Organization for Animal Health (WOAH) dashboard [29], which provided summarized data for reporting purposes. To estimate the recruitment rate through birth and the total population, we relied on data obtained from the Ethiopian Central Statistical Agency (CSA) for the year 2021/2022 [30]. The primary objective of this study is to determine the optimal parameters that allow our model (7) to be best aligned with the available data. To achieve this, we employ the nonlinear least-squares approximation method, which aims to minimize the sum of squared errors. where Ij is the prevalence data and I⁎(tj)=E(t)+I(t) is the simulation result (predicted) at time step j of prevalence of PPR. We used the initial value (S0,V0,E0,I0,B0,R0)=(14063965,3000000,10,684,10000,100), while all estimated parameter values are listed in Table 2 and the visualization of the result is shown in Fig. 3a. The blue curve represents the simulation result of I⁎(t) from the model (7), while the red dot is monthly prevalence data of PPR disease from January 2019 to Aug. 2022. Fig. 3b illustrates the residuals between the predicted and observed data. Comparison between the monthly prevalence of PPR disease data and simulation I⁎(t) with their residual. Using the parameters in Table 2, we estimated the control reproduction number (Rv) for PPR in the Amhara region of Ethiopia, producing a value of 1.0518. Under the assumption of zero influx of infectious restocked small ruminants and a newborn vaccination rate of 16.8%, the analysis predicts that PPR will remain an endemic disease in the Amhara region in the absence of better management efforts. Sensitivity analysis In order to assess the model's resistance to different parameter values, a sensitivity analysis was conducted. Understanding the relative roles played by various model parameters in the occurrence and spread of a given disease is vital. This understanding helps us identify where interventions should be focused to reduce disease-related mortality and morbidity. Several parameters contribute to (Rv) in the case of small ruminants with zero restocked infectious (α2=α3=0), as illustrated in equation (12). Consequently, a sensitivity analysis was performed in (Rv) to identify key variables that can limit the spread of PPR. To measure the relative change in the model parameter p relative to the relative change in (Rv), a forward normalized sensitivity index denoted γpRv was used. This index is defined using partial derivatives if (Rv) is a differentiable function of the model parameter p. The specific definition of this index can be found in [31]. It can be expressed as: where γpRv is the sensitivity index (SI) of (Rv) with respect to parameter p, hence we evaluate the sensitivity indices of these parameters at the parameter values as given in Table 3. For example, using the definition in (35), the sensitivity index of Rv with respect to β is given by This can be interpreted as increasing (or decreasing) the transmission rate β by 10%, increasing (or decreasing) Rv by 10%. In the same way, increasing (or decreasing) vaccination efficacy (1−ϵ), decreases (or increases) Rv, by 7.9%. In general, the positive sign sensitivity indices in Table 3 imply that the value of Rv increases with increase, whereas the negative sign sensitivity indices show that the value of Rv decreases with increase. Table 3Sensitivity indices of Rv for the PPR model (7) with (α2 = α3 = 0).Table 3Parameters:βμσγα1ρϵδκωIndex sign:+-+-+-+-+-SI values:1.0001.05860.08000.57400.04540.02700.79640.36230.05390.1304 Sensitivity indices of Rv for the PPR model (7) with (α2 = α3 = 0). Based on the results of the sensitivity analysis presented in Table 3, it is evident that enhancing the efficacy of the vaccine, reducing the transmission rate, and mitigating susceptibility resulting from restocking are all effective measures in preventing the spread of PPR disease. Furthermore, this finding is supported and verified in the numerical simulation section. Numerical simulation In this section, we examined the effects of parameter values on the dynamics of PPR and validated the analytical results using numerical methods. For numerical simulations, we used the parameter values specified in Table 2. Effects of vaccination (ρ and ω) on the dynamics of PPR Given that the efficacy of the vaccine at both the regional and national levels falls within the range of 60−80% [2], [3], or equivalently, ϵ lies within the range of 0.2−0.4. In particular ϵ=0.35, we can determine the critical value related to the vaccine from Remark 1(i) as ϵc=1/R0=0.3326 (that is, ϵ>ϵc). Fig. 4a presents a 3D plot of Rv in the ρω-Rv plane, consistently positioned above the plane Rv=1. This observation indicates that Rv never falls below 1. Consequently, despite the high levels of vaccination considered and the absence of infectious restocking of small ruminants, the PPR disease will persist in the small ruminant population. 3D-plots to show the effects of ω,ρ,ϵ on Rv. Furthermore, for ϵ=0.2(<ϵc), it is observed that the trajectory of Rv intersects the plane Rv=1. This can be seen in Fig. 4b. The simulation illustrates that with a combination of high efficacy and coverage of vaccination, the disease will eventually decline within the population (i.e., Rv<1). Therefore, this numerical result provides empirical validation for the influence of ϵ, as demonstrated by the sensitivity index. Effects of transmission rate (β) and fraction of restocked susceptible population (α1) We present a discussion on the influence of β and α1 on the dynamics of the infectious class I, as depicted in Fig. 5a. Through an examination of various values for β and α1, we demonstrate the impact on the reproduction number. The figure clearly illustrates that an increase in both α1 and β corresponds to an increase in the reproduction number, with the trajectory of Rv intersecting the plane Rv=1. Notably, at (α1=0.01, β=2.0e−7), the trajectory coincides with Rv=1, indicating that the disease will cease to propagate when β<2.0e−7, while it will persist if β>2.0e−7. Additionally, we have substantiated this result through a time series plot of the infection class in Fig. 5b. The analysis of forward bifurcation reveals that even a minor change or increment in the transmission rate can lead to significant alterations in the dynamics of the disease. Effects of fraction of susceptible restocked animal (α1) and transmission rate on the dynamics of PPR in the case of α2 = α3 = 0. Interplay between restocking and vaccination in the PPR dynamics To support the analytic results, we consider the parameter values presented in Table 2. We investigated the effects of values of ϵ in the range; ϵ=(0.1,0.4), while considering different initial conditions (IC) for the model. The IC values are as follows: In Fig. 6a, we observe that the trajectory of the infectious variable I in the time series plot converges to the endemic equilibrium I⁎=1.338×105 with a vaccination efficacy of 1−ϵ=0.65, assuming the absence of infection inflow from the restocked class. This result verifies the findings of Remark 1(i) and Theorem 4.3. Moreover, when there is less loss of immunity, that is, efficiency 1−ϵ=0.8 (80%), the dynamics of the disease change significantly and the trajectory of I converges to the disease-free equilibrium (I=0), as shown in Fig. 6b. Simulation results to show effects of ϵ and α on I. Furthermore, in Fig. 6c, we notice the presence of infective inflow through restocking (i.e., α1=0.01,α2,3=0.01) results in the existence of a globally stable endemic state I¯=1634, even when high efficacy vaccination (ϵ=0.1) and a vaccination coverage rate of ω=56.9% of the susceptible population are applied. This result indicates that the introduction of infective individuals through restocking facilitates the persistence of the PPR virus in the region. Therefore, additional control mechanisms must be implemented alongside vaccination. However, vaccination campaigns should be strengthened with high efficiency to reduce the burden of PPR in the region. In Fig. 6d, it can be observed that there is a high flow of small infectious ruminants. This is indicated by the fact that ω>ωc and b¯>0, as derived from the result presented in Theorem 4.5 (ii). As a consequence, all the state variables in the model (11) converge to the endemic equilibrium ξ¯=(8.562×105,8.616×106,4.320×105,3.990×105,3.605×106,3.442×105). This outcome of the simulation is consistent with the findings stated in Theorem 4.5. Fig. 6e illustrates the impact of restocking on the dynamics of PPR by considering different combinations of αi values, where i=1,2,3. It can be observed that as the total fraction of infective restocked population increases, the infection population also increases within the small ruminant population. This emphasizes the significance of restocking in influencing the spread of PPR. In Fig. 7, we present a 3x3 subplot layout to examine the interaction between vaccination efficacy and restocking dynamics during disease outbreaks. Each column focuses on the effects of restocking parameters (α1,α2,α3), with each row representing increasing levels of vaccination efficacy. Each subfigures displays varying restocking levels. In the first column, the subfigures (7(a), 7(b) and 7(c)) illustrate the effect of α1. Results indicate that higher values of α1 increase the infection peak, as more susceptible ruminants are available for the disease to spread, particularly when vaccination efficacy is low. However, with high vaccination efficacy, the impact of α1 is reduced. The effects of α2 and α3 are illustrated in the subfigures located in the second and third columns, respectively. The infection peaks are notably higher with increased values of α2 and α3, as demonstrated in Figures (7d-7i), thereby extending the duration of the outbreak. Although high vaccination efficacy reduces the number of infections, it cannot completely eliminate the disease. This underscores the need for additional interventions to control outbreaks due to the restocking of infected animals. Interplay between restocking and vaccination in PPR disease. Numerical verification of global stability of DFE and EEP As indicated by the analytical results presented in Theorem 4.2, Theorem 4.6, the global stability of both the disease-free equilibrium (DFE) and the endemic equilibrium points (EEP) offers valuable insights into the long-term behavior of PPR within the small ruminant population. It informs us whether the disease can be effectively controlled or is likely to persist and become endemic. A comprehensive understanding of the global stability of these equilibrium points plays a crucial role in interpreting the potential impact of interventions, evaluating the dynamics of disease transmission, and guiding strategies for disease prevention and control on a global scale. Considering various initial conditions specified in (36), numerical verification can be used to establish the global stability of disease-free equilibrium (DFE) and endemic equilibrium points (EEP) for the two models (7) and (11). In Fig. 8a, we present the dynamic trajectory of the model (11) with specific parameter values: ϵ=0.3, α1=0.9 and the remaining parameters listed in Table 2. The figure shows that the system (11) exhibits an endemic equilibrium ξ⁎=(S⁎,V⁎,E⁎,I⁎,B⁎,R⁎)=(1.106×106,1.066×107,3.623×105,2.651×105,2.358×105,2.393×106), which is globally asymptotically stable. Similarly, Fig. 8b illustrates the behavior of model (11) with the same parameter values, except for ϵ which is set to ϵ=0.15. In this case, the model exhibits a disease-free equilibrium ξ0=(1.205×106,1.509×107,0,0,2.358×105,0), which is globally asymptotically stable for different initial values in (36). These results provide empirical support for the analytical findings presented in Theorem 4.2. Verification of global stability using different initial conditions in (36). Furthermore, Fig. 8c indicates that the trajectories of the state variables in model (7) converge to a unique endemic equilibrium point ξ¯=(8.597×105,8.567×106,4.426×105,4.094×105,2.219×105,3.695×106) where ϵ=0.2 and αi=0.32,i=1,2,3. Moreover, with the initial conditions specified in (36), this equilibrium point ξ¯ is globally asymptotically stable. This numerical verification reinforces the result stated in Theorem 4.6. Discussion and conclusion Peste des petits ruminants (PPR) is a significant endemic disease of small ruminants in Ethiopia, caused by the PPR virus. Despite the availability of a lifetime immunity-providing vaccine, PPR outbreaks continue to spread in Asia and Africa, indicating that control and eradication efforts can be improved by considering various risk factors. In this study, we incorporated a restocked group into a deterministic SEVIR epidemic model to assess the impact of restocking on PPR dynamics. This model, like other epidemiological models of infectious animal diseases, allows us to simulate disease spread and evaluate control strategies effectively. Our analytical and numerical results revealed critical insights into the persistence of PPR in the Amhara region. In the scenario without the influx of infective animals due to restocking, we observed that the basic reproduction number (Rv), estimated at 1.0518, indicated the endemicity of the disease in the region. When Rv<1, the disease-free equilibrium was globally asymptotically stable, while the endemic equilibrium was locally stable when Rv>1. In numerical simulation, we demonstrate that high vaccine efficacy could potentially lead to disease control or eradication without the complicating factor of restocking dynamics. In contrast to the scenario that included restocking, we investigated the existence and global asymptotic stability of a unique endemic equilibrium point. This scenario highlighted that the influx of susceptible and infected animals from surrounding regions has a significant impact on disease dynamics. Even with ideal coverage and efficacy of the vaccine, the introduction of PPRV through restocking could maintain the disease within the population, emphasizing that vaccination efforts alone may not be sufficient to control the outbreak. The findings suggest that while improving vaccination efficacy and coverage is crucial, external influx of infection due to restocking poses a substantial challenge to controlling PPR. This aligns with previous studies [3], which indicate that the Amhara region, characterized as a highland or sedentary agro-ecological zone, interacts with adjacent lowland areas that serve as reservoirs for the PPR virus. Socioeconomic practices that facilitate animal movements between these areas further complicate disease control efforts. Our results advocate for enhanced biosecurity measures and prerestocking evaluations to mitigate the risks associated with the introduction of infected animals. Furthermore, evidence from previous statistical seroprevalence studies [6], [2] suggests the circulation of Peste des Petits Ruminants (PPR) in the region. Despite differences in methodological techniques, both investigations support our findings on disease epidemiology, associated risk factors, and prevention strategies. Our mathematical modeling approach improves this understanding by predicting disease trends and simulating the intricate interaction of vaccination efforts and risk factors such as restocking. This comprehensive paradigm helps to provide an expanded perspective of PPR transmission and control within the current epidemiological picture capable of keeping the outbreak under control. Despite these findings, our study has limitations, notably in terms of the availability of relevant data. Future studies should include a broader spectrum of risk factors that influence PPR dynamics, as well as strong data sets for more extensive modeling. Furthermore, investigating optimal control problems using different intervention strategies will be critical for efficient disease management. In conclusion, effective strategies are critical for achieving herd immunity and strengthening PPR control measures in the region. By implementing strong vaccination programs, completing rigorous pre-restocking evaluations, and improving biosecurity measures, we can considerably limit the impact of restocking on PPR dynamics. Future research should investigate a larger range of risk factors and effective control measures to further eradicate PPR and ensure the health of small ruminant populations in the region and at the national level. Funding No funding was received for conducting this study. CRediT authorship contribution statement Yibekal Walle: Writing – review & editing, Writing – original draft, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Joseph Y.T. Mugisha: Writing – review & editing, Supervision, Methodology, Conceptualization. Dawit Melese: Writing – review & editing, Validation, Supervision, Methodology, Conceptualization. Haileyesus Tessema: Writing – review & editing, Writing – original draft, Validation, Supervision, Software, Conceptualization. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Contributor Information Yibekal Walle, Email: yibekalw@gmail.com. Joseph Y.T. Mugisha, Email: joseph.mugisha@mak.ac.ug. Dawit Melese, Email: dawitmpg@gmai.com. Haileyesus Tessema, Email: haila.tessema@gmail.com. Data availability The data that support the findings of this study are openly available in the World Organization for Animal Health (WAHIS) at https://wahis.woah.org/#/dashboards/qd-dashboard, and monthly reported data taken from the Ministry of Agriculture of Ethiopia. References Associated Data Data Availability Statement The data that support the findings of this study are openly available in the World Organization for Animal Health (WAHIS) at https://wahis.woah.org/#/dashboards/qd-dashboard, and monthly reported data taken from the Ministry of Agriculture of Ethiopia.