Impulsive intervention strategies for temperature and rainfall-dependent visceral leishmaniasis transmission dynamics Authors: Temesgen Debas Awoke, Semu Mitiku Kassa, Gizaw Mengistu Tsidu Published: 2026 ABSTRACT AbstractVisceral leishmaniasis (VL) remains a lethal parasitic disease, disproportionately affecting resource-limited regions where sustained control measures are often economically and logistically impractical. Consequently, cost-effective and sustainable strategies tailored to seasonal transmission patterns are urgently needed. Because temperature and rainfall strongly influence sandfly populations, aligning control efforts with seasonal transmission dynamics may enhance effectiveness while reducing costs. This study proposes an impulsive control strategy within a temperature- and rainfall-dependent VL transmission model to assess how strategically timed, short-term interventions optimize disease control. We examine the effects of intervention timing, frequency, and coverage for measures including sandfly breeding site elimination, insecticide spraying, and culling infected reservoir animals. Theoretical analysis shows that the disease-free periodic solution is locally asymptotically stable when the basic reproduction number () is below one, while endemic persistence occurs when . The calibrated model closely reproduced observed seasonal transmission patterns, providing a robust basis for evaluating interventions under climatic forcing. Global sensitivity analysis revealed that vector-related parameters consistently drive infection burden, while reservoir parameters show negligible effects, indicating that vector control should be prioritized over reservoir culling for cost-effective VL management. Simulations indicate that targeted interventions implemented for only a few weeks annually substantially reduce transmission. Although vector control and reservoir culling independently decrease cases, their combined application is more effective. A biannual one-week intervention reduces human cases by 95.11%, increasing to 96.6% when extended to two weeks. Initiating interventions six weeks after peak infection yields the most substantial long-term impact, achieving a 98.89% reduction at 85% coverage.Keywords: Visceral leishmaniasis, Impulsive control, Climate variable-dependent VL, Reservoir animal culling, Vector controlSubject terms: Diseases, Ecology, Ecology, Mathematics and computing Introduction Visceral leishmaniasis (VL), commonly referred to as kala-azar, is one of the most severe parasitic diseases, next to malaria, and always fatal if left untreated1. It poses a significant global health challenge, affecting millions of people worldwide and disproportionately impacting vulnerable populations in endemic regions. An estimated 50,000 to 90,000 new VL cases occur worldwide annually, with only 25-45% reported to WHO1. Ninety percent of reported VL cases occur in Brazil, Ethiopia, India, Kenya, Somalia, South Sudan, and Sudan2. Leishmania parasites are transmitted to humans through the bites of infected female phlebotomine sandflies, which require blood meals for egg production. The complexity of the disease transmission is exacerbated by environmental factors such as rainfall patterns, temperature, soil types, and vegetation which significantly influence the life cycle and distribution of both the sandfly vectors and the Leishmania parasites3. Rainfall patterns affect breeding sites for sandflies, while temperature plays a critical role in the development and survival of the parasite within the vector. The number of VL cases rises during the dry season and declines during the rainy season in a recurring pattern4. Since Dye et al.5 introduced the first classical SIR model for VL transmission dynamics, numerous mathematical models have been developed and extensively analyzed, significantly enhancing the understanding and management of the disease2,6–9. For instance, Awoke et al.2 developed a sex-structured disease transmission model to investigate the dynamics and control of VL. The study uniquely incorporated sex-specific parameters, recognizing that males are disproportionately affected by the disease. The model accounted for key components such as human, sandfly vector, and animal reservoir populations, emphasizing the impact of sex differences on transmission dynamics. Control mechanisms, including insecticide-treated bed nets, vector control interventions, and treatment strategies, were analyzed for their efficacy in reducing disease burden. The authors conducted numerical simulations to explore the impact of these interventions, highlighting the importance of targeted control strategies tailored to the epidemiological and demographic characteristics of VL. Gumaa et at.8 present a comprehensive mathematical modeling study analyzing the impact of various control strategies on the dynamics of VL. The authors developed a compartmental model to capture the transmission dynamics of VL, incorporating human, vector, and reservoir populations. The model examines the effectiveness of interventions such as insecticide-treated nets, vector control, and treatment strategies. Through numerical simulations, the study evaluates the impact of these control measures on reducing disease prevalence and highlights the significance of integrated approaches for sustainable VL control. The findings underscore the importance of timely and targeted interventions to achieve significant reductions in disease burden. Khatua et al.9 provided an insightful perspective on the integration of mathematical modeling and machine learning techniques to predict and understand VL dynamics. The paper highlights the strengths of mathematical models in simulating the transmission dynamics of VL, considering biological and environmental factors, and their potential for informing control strategies. Furthermore, the authors explore the emerging role of machine learning techniques in identifying patterns and predicting disease trends using large data sets, such as climate variables, vector population dynamics, and case incidences. By combining these approaches, they advocate for a synergistic framework that leverages the explanatory power of mathematical models with the predictive capabilities of machine learning. Davis et al.7 developed a mathematical model to investigate the transmission dynamics and control strategies for VL in the Indian subcontinent, with a particular focus on the impact of insecticide-treated nets (ITNs). The study incorporated key epidemiological factors, including human and sandfly populations, to simulate VL transmission and assess intervention outcomes. Their model demonstrated the effectiveness of ITNs in reducing VL incidence by minimizing human-vector contact and interrupting transmission pathways. The authors also explored the conditions required for VL elimination, highlighting the importance of sustained ITN coverage alongside other control measures. Recent studies have advanced the mathematical modeling of VL, offering diverse perspectives on disease transmission dynamics, control strategies, and environmental influences. Modu et al.10 proposed a deterministic model for zoonotic VL that incorporated human, vector, and canine populations, providing insights into transmission dynamics and intervention efficacy. El Gumaa et al.8 extended this approach by evaluating the effects of various control strategies-such as treatment, vector control, and awareness campaigns,on disease prevalence, emphasizing their potential to reduce infection burden. Elmojtaba et al.11 developed an optimal control analysis of an VL-HIV co-infection model, highlighting the synergistic effects of prevention and treatment interventions while underscoring the complexities introduced by co-morbidity. The current control strategies for VL rely more on reservoir and vector control strategies together with the use of insecticide-impregnated materials and active case detection and treatment12. Stauch et al.14 explored the efficacy of different vector-based control measures, and they have found that reducing the sandfly population by 79% via reduction of breeding sites, or reducing the sandfly population by 67% through increasing sandfly mortality, are both sufficient to eliminate the Leishmania donovani parasite in the human population. One of the challenges of controlling VL is due to the role of reservoirs. To address these challenges, researchers have explored various intervention strategies, including the culling of reservoir animals. The inclusion of culling infected dogs as a strategy for controlling visceral leishmaniasis has been integrated into the models in the references13,15–17. For example, Sevá et al.17 explored canine-based strategies for the prevention and control of VL in Brazil, focusing on the role of domestic dogs as primary reservoirs. The study evaluated the effectiveness of interventions such as dog culling, insecticide-treated collars, and vaccination in reducing disease transmission. The authors demonstrated that combining multiple strategies yields better outcomes than any single approach, highlighting the importance of integrating vector control with canine-focused measures. Their findings emphasize the critical role of dogs in VL transmission dynamics and underscore the need for a holistic approach to disease management in endemic regions. Quinnell and Courtenay16 also provided an in-depth review of the transmission dynamics, reservoir hosts, and control strategies for zoonotic VL. The authors discussed the role of domestic and wild dogs as primary reservoirs, highlighting their contribution to the persistence of the disease in endemic areas. The paper also examines various control strategies, such as vector control, reservoir host management, and vaccination, and underscores the need for integrated approaches to effectively reduce VL transmission and mitigate its impact. Costa et al.18 highlighted the impact of culling strategies for controlling VL transmission in dogs. The authors showed that screening 4% of symptomatic dogs monthly, with a 4-month delay before culling, effectively reduces transmission to less than 1% in low-transmission areas. However, in high-endemic regions, at least 30% of asymptomatic but infectious dogs would need to be culled for similar results. Palatnik-de-Souza et al.19 and Ribas et al.20 analyzed significantly higher culling rates, assuming uniform infectiousness among dogs. Palatnik-de-Souza et al. determined that removing 7% of infectious dogs daily was essential to reduce the basic reproduction number () below the critical threshold of unity. Similarly, Ribas et al.20 found that culling less than 18% of dogs daily would have minimal impact on controlling disease transmission to humans. Ming et al.21 developed a vector-borne disease model that integrates both time delays and impulsive control strategies to capture the dynamics of periodic vector interventions. Their model considers the transmission between human and vector populations, where the impulses represent discrete control measures such as vector culling or insecticide spraying applied at specific time intervals. The authors supported their theoretical findings with numerical simulations showing how varying the impulsive control intensity and delay length influences disease prevalence. Despite the model’s mathematical rigor and insightful analysis, it presents several limitations. In particular, the absence of a reservoir population restricts its applicability to human-only transmission systems, overlooking the critical role that reservoir animal hosts play in maintaining infection cycles. Moreover, the model assumes homogeneous mixing between humans and vectors and omits environmental or seasonal factors that often drive vector population dynamics. The impulsive control is restricted to the vector population alone, without consideration of host- or reservoir-targeted interventions, and is applied at fixed time intervals rather than optimized campaign schedules. Consequently, while the work of Ming and Yu provides a valuable theoretical foundation for understanding impulsive control in vector-borne diseases, it remains limited in scope for addressing multi-host, non-autonomous transmission systems. VL cases tends to exhibit a seasonal pattern, with a notable increase during the dry season and a decline during the rainy season4. During the dry season, sandfly activity often intensifies due to favorable breeding and feeding conditions, leading to a higher rate of human-vector contact and increased transmission. Conversely, the rainy season can disrupt sandfly habitats by flooding breeding sites and altering their ecological dynamics, leading to a temporary reduction in transmission. These seasonal variations highlight the critical role of environmental factors in shaping the epidemiology of VL, emphasizing the need for tailored intervention strategies that align with these patterns to effectively control disease transmission. Awoke et al.2 developed a sex-structured disease transmission model for VL, incorporating control mechanisms such as treatment, vector control, and reservoir management. Their study highlights the differential exposure and disease burden among males and females. It underscores the importance of incorporating heterogeneity into disease modeling to improve intervention planning and cost-effectiveness. However, existing VL control measures are assumed to typically be implemented continuously throughout the year, which poses substantial technical, logistical, and economic challenges, especially in resource-limited settings. Sustaining continuous interventions over extended periods is often financially unfeasible, requiring a significant allocation of resources that may not be readily available. Furthermore, maintaining high levels of community engagement and operational efficiency over prolonged intervention periods is difficult especially in rural areas, leading to less practical and less efficient disease control outcomes22. Given these challenges, a more cost-effective and practical approach is to implement VL control measures at critical times of the year through a short-term, intensive campaign rather than continuous application. Targeted interventions such as the destruction of sandfly breeding sites, indoor and outdoor residual spraying (IRS) of houses and animal shelters, and the culling of infected reservoir animals can be strategically timed to coincide with critical transmission periods. Implementing these interventions in concentrated efforts for one, two, or more weeks during high risk periods may potentially maximize their impact while minimizing costs and resource expenditure. This campaign-based approach may ensure that VL control efforts if epidemiologically effective, could be economically sustainable, optimizing resource allocation while significantly reducing the burden of the disease. Despite the widespread application of impulsive control strategies in mathematical modeling of vector, borne diseases such as malaria38–40, West Nile virus37, and dengue36, no study to date has developed or evaluated impulsive control frameworks specifically tailored to VL transmission under climate-driven seasonal forcing. While impulsive interventions, including periodic vector spraying, insecticide-treated nets, and mass drug administration, have demonstrated measurable effectiveness in reducing transmission for other diseases, the epidemiological and ecological characteristics of VL differ fundamentally from these diseases. Unlike malaria, where transmission is primarily driven by direct mosquito-human contact, VL involves complex zoonotic reservoirs (e.g., dogs, rodents), a prolonged asymptomatic human incubation period lasting months to years, and sandfly vector biology that exhibits distinct and nonlinear sensitivity to temperature and rainfall41–43. Moreover, the optimal timing, frequency, and duration of impulsive controls for VL remain unexplored in the existing literature. Consequently, key questions regarding when and how often interventions should be applied to maximize impact while minimizing resource use have yet to be answered. To address this critical gap, the present study develops a novel impulsive control framework for VL that integrates empirically-informed, climate-dependent vector parameters and systematically evaluates intervention timing strategies using a formal objective function. This tailored approach provides the first quantitative evidence for optimizing impulsive control measures specifically for VL transmission in resource-limited, seasonally driven endemic settings, thereby offering actionable insights for public health planning and intervention design. Motivated by this approach, we extend the VL model developed by Awoke et al.2 to an impulsive control model. Since medical treatment is administered according to its prescribed schedule rather than as a short-term campaign, the authors didn’t consider it in this model. This model focuses on optimizing the timing, frequency, and intensity of interventions to enhance cost-effectiveness and long-term disease reduction. The study systematically identifies critical intervention initiation times. Additionally, we assess the impact of varying intervention durations, examining how extending or shortening the application period influences disease suppression. Moreover, we explore the role of intervention frequency, determining whether periodic once-per-year or twice-per-year interventions yield greater impact on reducing the prevalence of the disease. By integrating these key factors, this research provides practical, evidence-based recommendations for optimizing VL control programs. The rest of the paper is organized as follows. In Sect. Model formulation, we formulate the VL model with impulsive control. In Sect. Mathematical analysis, the stability of the disease-free solution and permanence of the model are proved. In Sect. Numerical simulation and discussion, numerical simulations are conducted, accompanied by a detailed discussion. Finally, the conclusions are addressed in Sect. Conclusion and recommendation. Model formulation In this work, we formulate and study a deterministic mathematical model for the transmission dynamics of temperature and rainfall-dependent VL by extending the VL model studied in Awoke et al.2 into the non-autonomous system of equations to incorporate impulsive intervention strategies. The host population is subdivided into three compartments: The total human population at time t denoted by , is divided into the following epidemiological subgroups: , Susceptible human individuals; , Exposed human individuals; , infected individuals who have active VL and are infectious and , effectively recovered individuals from the disease. The total sandfly population at time t () is also divided into: , Susceptible sandfly population and , infected sandfly population. Similarly, the total reservoir population at time t () is divided into: , Susceptible reservoir individual and , infected reservoir individual. Hence, , and . The susceptible human population cohort increases by a constant recruitment, , as well as by the loss of immunity from recovered class at a rate θ. This population decreases by a temperature dependent infection at a rate and natural mortality at a rate , where , with a temperature dependent biting rate α (T) and representing the probability that a susceptible human becomes infected through a single bite of an infected sandfly. The size of Exposed individuals increase through infection from the susceptible class at a rate . It decreases by natural mortality at a rate and by becoming infectious at a rate τ. The population of infectious human is increased through developing infections from the exposed individuals at a rate τ, while it diminishes by recovery from the infection, disease induced death and natural mortality at rates γ, and respectively. The population of the recovered humans increases by the healing of infected individuals at a rate γ. This group decreases by natural mortality and by loss of immunity and becoming susceptible again at the rates and θ respectively. The susceptible sandfly population increases by a recruitment, . It decreases by natural death at a rate , and get infected at a rate , where and represent the probability that a susceptible sandfly becomes infected when feeding on the blood of an infected human, and reservoirs respectively. Infected sandflies increase by infection from the susceptible vector by a rate and decrease by natural death rate . The susceptible reservoir population also increases by a recruitment rate and decreases due to infection at a rate and natural mortality rate , where represents the probability that a susceptible reservoir becomes infected through a single bite from an infected sandfly. Infected reservoir individuals increase by infection from the susceptible vector at a rate and decrease by natural death rate . Temperature, T(t) and rainfall, Υ (t) values are assumed to be continuous, bounded, positive and ω-periodic functions. Furthermore, , α (T) and are also assumed to be continuous, bounded, positive and ω-periodic functions of the time variable t. To investigate the management of impulsive vector control, we consider the strategy of implementing destruction of vector breeding sites and insecticide residual spraying (IRS) at some fixed times as well as culling of infected animals at the specific critical times, using impulsive differential equations, which leads to the following model (see Table 1 for the description of the variables and parameters of the system (1)): where ω is a fixed positive constant and denotes the period of the impulsive effect, p∈ (0,1) represents the fraction of sandflies eliminated through the destruction of breeding sites at the critical times nω; q∈ (0,1) indicates the proportion of sandflies killed using an insecticide residual spraying strategy at these critical times; and m∈ (0,1) signifies the proportion of infected animals or reservoirs removed as part of an animal culling strategy at the critical times nω. Here, and represent the number of susceptible and infected sandflies immediately after vector control interventions at time nω, and is the number of infected animals right after the culling of infected reservoir animals. Table 1Description of the variables and parameters of the system (1).VariableDescription, , Susceptible human, sandfly, and reservoir populationsExposed human population, , Infected human, sandfly, and reservoir populationsRecovered human populationParameterDescription, Recruitment rate of human, sandfly and reservoir populationsLife expectancy of human, sandfly and reservoir populationsInfection rate of human, sandfly and reservoir populationsDisease-induced death rates of human and reservoir populationsτInverse of incubation period of human populationθRate of loss of infection acquired immunity of human populationγRecovery rate of human population, Probability that a susceptible human and reservoirs become infected through a single bite of an infected sandfly, Probability that a susceptible sandfly becomes infected when feeding on an infected human, and reservoirsαNumber of bites in which the host receives per week Description of the variables and parameters of the system (1). In this model, it is assumed that the intervention period is synchronized with the climatic seasonal period, i.e both occur at the same frequency and have equal duration. This implies that interventions are applied periodically in alignment with the seasonal cycles, ensuring that control measures coincide with key environmental and epidemiological changes. This synchronization allows for a more effective assessment of how interventions interact with seasonal variations in temperature, rainfall, and vector dynamics. By structuring interventions to match the natural seasonality of VL transmission, we aim to optimize their impact and capture the realistic timing of vector control efforts in endemic regions. Temperature and rainfall are assumed to exhibit periodic behavior with a one-year cycle (equivalent to 52 weeks). Then, considering the temperature and precipitation data obtained from NASA POWER23 for the geographic points in Northwestern Ethiopia. Hence, the corresponding temperature-dependent infection rates, vector recruitment rate, and biting rate, as well as temperature and rainfall-dependent sandfly mortality rate, become where the modification functions and are given in Eqs. (2) and (3) and see Appendix A for the detailed. According to the findings reported in Kasap et al.24, sandfly mortality significantly increases when temperatures exceed or fall below . This suggests that the optimal temperature range for sandfly survival and reproduction lies between and . Within this interval, the environmental conditions are favorable, allowing sand flies to thrive. However, when temperatures go beyond this threshold, either too high or too low, the mortality rate of sand flies rises sharply, hindering their population growth. Based on this observation, we introduce a temperature-dependent modification parameter, denoted as , to reflect the influence of temperature on sandfly dynamics. Furthermore, environmental factors such as heavy rainfall play a crucial role in sandfly development. Specifically, excessive rainfall can inundate breeding sites, particularly those harboring sand fly larvae, thereby disrupting their life cycle. In such scenarios, typically associated with colder and wetter periods, the sand fly population tends to decline due to unfavorable conditions. To incorporate the combined effects of rainfall and temperature on sand fly ecology, and considering that weekly rainfall values exceeding 7 mm are detrimental to larval habitats in the study area, we define a piecewise rainfall-dependent modification parameter, denoted as , as follows: This article also incorporates the following key assumptions into the model framework: The biting rate of sandflies depends predominantly on air temperature. The biting behavior of sandflies is strongly influenced by ambient temperature and plays a crucial role in the transmission dynamics of VL. Sandfly activity, particularly host-seeking and feeding patterns, is highly temperature-sensitive. At optimal temperature ranges, their metabolic processes are stimulated, leading to more frequent biting. However, when temperatures fall outside this range, either too low or excessively high, sandfly activity declines due to physiological stress or dormancy, resulting in reduced biting rates. Consequently, integrating a temperature-dependent biting rate into transmission models captures the ecological and behavioral variability of sandflies, enhancing the accuracy and realism of disease dynamics predictions. The biting rate of sandflies depends predominantly on air temperature. The biting behavior of sandflies is strongly influenced by ambient temperature and plays a crucial role in the transmission dynamics of VL. Sandfly activity, particularly host-seeking and feeding patterns, is highly temperature-sensitive. At optimal temperature ranges, their metabolic processes are stimulated, leading to more frequent biting. However, when temperatures fall outside this range, either too low or excessively high, sandfly activity declines due to physiological stress or dormancy, resulting in reduced biting rates. Consequently, integrating a temperature-dependent biting rate into transmission models captures the ecological and behavioral variability of sandflies, enhancing the accuracy and realism of disease dynamics predictions. The recruitment rate of sandflies is temperature dependent. Temperature significantly influences the reproductive biology of sandflies, affecting their growth, survival, and reproductive output. Within an optimal temperature range, warmer conditions speed up larval development and reduce the time needed for eggs to mature, resulting in increased recruitment rates. Conversely, exposure to temperatures outside this optimal range can hinder larval growth or elevate mortality, thereby lowering recruitment. Incorporating a temperature-dependent recruitment rate into the model allows it to reflect climate-driven fluctuations in sandfly population dynamics, consistent with biological evidence and field data. The recruitment rate of sandflies is temperature dependent. Temperature significantly influences the reproductive biology of sandflies, affecting their growth, survival, and reproductive output. Within an optimal temperature range, warmer conditions speed up larval development and reduce the time needed for eggs to mature, resulting in increased recruitment rates. Conversely, exposure to temperatures outside this optimal range can hinder larval growth or elevate mortality, thereby lowering recruitment. Incorporating a temperature-dependent recruitment rate into the model allows it to reflect climate-driven fluctuations in sandfly population dynamics, consistent with biological evidence and field data. The natural death rate of sandflies is both temperature and rainfall-dependent. Sandflies are highly responsive to environmental factors, with their survival closely linked to variations in temperature and rainfall. Extreme temperatures can lead to direct mortality due to thermal stress, while rainfall influences the availability of breeding and resting habitats. Heavy rainfall can wash away breeding sites, whereas too little rainfall may cause habitat desiccation, increasing mortality rates. By integrating temperature and rainfall dependencies into the natural death rate, the model captures the impact of these climatic variables, providing a more accurate and realistic depiction of sandfly population dynamics under changing environmental conditions. The natural death rate of sandflies is both temperature and rainfall-dependent. Sandflies are highly responsive to environmental factors, with their survival closely linked to variations in temperature and rainfall. Extreme temperatures can lead to direct mortality due to thermal stress, while rainfall influences the availability of breeding and resting habitats. Heavy rainfall can wash away breeding sites, whereas too little rainfall may cause habitat desiccation, increasing mortality rates. By integrating temperature and rainfall dependencies into the natural death rate, the model captures the impact of these climatic variables, providing a more accurate and realistic depiction of sandfly population dynamics under changing environmental conditions. Temperature, T(t) and rainfall, Υ (t) are assumed to be continuous, bounded, positive and ω-periodic functions. Temperature, T(t) and rainfall, Υ (t) are assumed to be continuous, bounded, positive and ω-periodic functions. , α (T) and are also assumed to be continuous, bounded, positive and ω-periodic functions. , α (T) and are also assumed to be continuous, bounded, positive and ω-periodic functions. Notation: For convenience, we change the notations of , , Υ (t)) and α (T(t)) to and α (t) respectively. Mathematical analysis Analysis of the disease-free periodic solution(DFPS) Firstly, we analyze the existence and stability of the disease-free periodic solution for system (1) with , , , and , denoting as then system (1) can be reduced as follows We can see that humans and sandflies are independent of one another in this subsystem. For the human and reservoir populations, the impulsive controls have no effect on them and it can be obtained that as and as . On each interval , the ODE is given by The general solution, with initial condition , is At the impulsive time t = (n+1)ω, we have The periodic steady state satisfies Substituting into the continuous solution on gives Hence, for the sandfly equations, the periodic solution on every interval has the form: which is globally stable25,26 and the following conclusion holds true. Lemma 1 For system (6), there exists a unique positive ω periodic solution with , and as for every solution of (6). The proof follows a similar argument as in Zhang et al.27, hence omitted. Based on the results of Lemma 1, system (1) admits the DFPS on every impulsive interval . To determine the stability of DFPS of system (1), we define Let be an n× n matrix, be the fundamental solution matrix of the system of linear ordinary differential equations and be the spectral radius of . Then we use the following transformation to study the local stability of DFPS. , , , , , , , . Using these expressions and denoting (the order of the equations is adjusted for convenience), we linearize system (1) at DFPS, which leads to the following system. where and and can be expressed in the form of a block matrix as shown below: with Then solving the system leads to At the fixed times t=nω, it can be easily determined that It is evident that . Representing , then if is also less than one, we have the following result. Theorem 1 The disease-free periodic solution is locally asymptotically stable when . Persistence of the disease In this section, we examine the persistence of the disease, which implies that under certain conditions, the disease will continue to exist rather than be eradicated. To establish this result, we first present the following lemma. Lemma 2 When , there exists ψ > 0 such that the solution of system (1) satisfies Proof We derive it using a proof by contradiction. Assume that there exists a time such that , and when . Then according to the equations about , and in system (1), we consider the following system By the comparison principle, we deduce that Similar to system (6), it follows that system (8) admits a positive periodic solution , which is globally asymptotically stable, and where represent disease free periodic solution is valid for each j=h, v, r. Then there exists small enough and , such that , and for . This implies that Substituting the above inequalities into equations of , and in system (1) yields Here, from line two of Eq. (1), i.e , we have that . Hence, when . Next, we deal with the variables , and . In fact, adding the equations for human sub-populations, sandfly population and reservoir population respectively gives Representing as the solution of system (10) and as that of system (6) and comparing the equations in system (10) with that in system (6), it is easy to get , and . Noticing that , , and can be chosen arbitrarily, then we have , , , and system (9) can be modified as: By comparison principle, we have , , , where () is a solution of the following system Specifically, we have and . It follows that , and as when , and then , and , which is a contradiction with the boundedness of , and . This completes the proof. □ Model calibration and sensitivity analysis Reliable parameter estimation is a cornerstone of mathematical modeling in epidemiology, as it directly influences the predictive capability and practical utility of the model. The primary objective of the model calibration process is to estimate key unknown parameters that govern disease transmission, progression, and control, ensuring that the model output closely matches the observed epidemiological trends in the study area. The data used for model calibration is presented in Fig. 1 that presents the monthly time series data collected from Abdurafi, Ethiopia, spanning 66 months (January 2010 to June 2015). The dataset comprises three key variables: VL case incidence (Fig. 1a), mean monthly temperature (Fig. 1b), and mean monthly rainfall (Fig. 1c). VL cases exhibit notable fluctuations ranging from 0.07 to 1.62 cases per 1000 population, with a mean incidence of 0.72 ± 0.35 cases per 1000. The highest VL peaks occurred in December of 2010 (1.58 cases) and March of 2015 (1.62 cases), suggesting potential seasonal transmission patterns. Temperature data reveal a distinct seasonal cycle, with maxima reaching C during April-May and minima dropping to C in August, characteristic of the semi-arid climate of the Ethiopian lowlands. Rainfall demonstrates pronounced seasonality, with the majority of precipitation concentrated between June and September (the Kiremt rainy season), peaking at 268.95 mm in July of 2014. Notably, VL cases appear to lag behind rainfall peaks by approximately 2-3 months, a pattern consistent with the ecological dynamics of sandfly vectors, which require warm temperatures and moderate humidity for optimal breeding and survival. The prolonged dry periods (October-May) typically coincide with lower VL incidence, except for a notable outbreak in March of 2014, possibly linked to residual environmental moisture from the preceding rainy season. This dataset provides a robust foundation for parameterizing and validating mathematical models of VL transmission dynamics in the context of seasonal climatic forcing, enabling the estimation of key epidemiological parameters and the evaluation of potential intervention strategies in resource-limited settings. Time series of (a) Visceral Leishmaniasis (VL) cases (per 1000 population), (b) mean monthly temperature (° C), and (c) mean monthly rainfall (mm) in Abdurafi, Ethiopia, from January 2010 to June 2015 (66 months). The blue dashed line in panel (a) indicates the mean VL case incidence (0.73 cases per 1000). Data sources: VL cases obtained from4; temperature and rainfall data retrieved from NASA POWER climate archive (available at: https://power.larc.nasa.gov/data-access-viewer/). The total human population in the study region is individuals. Based on epidemiological assumptions and local disease prevalence reports, the initial conditions for the human compartments are distributed as follows: susceptible individuals (30% of the total population), exposed individuals (6.15%), infectious individuals (10%), and recovered individuals (14.3%). For the sandfly population, the initial conditions are set to (28.76% of the human population) and (4.42%), based on typical vector-to-host ratios reported in similar ecological settings. The reservoir animal population is initialized with (30.98%) and (6.64%). The estimated parameter values are presented in Table 2, while the corresponding model fit is shown in Fig. 2. Table 2Model parameters, units, values, and source.ParameterUnitRangeValueSourcepersons week−1–34.8379Estimatedreservoirs week−1–40.92Estimatedweek−1–0.0012129Estimatedweek−1–0.00693Estimatedγweek−1–0.10833Estimatedτweek−1–0.1111Estimatedθweek−1–0.0273Estimatedvectors week−1–VariableCalculateddimensionless–VariableCalculatedα (T)week−1–VariableCalculatedweek−1–VariableCalculatedproportion[0.01, 0.05]0.047187Fittedproportion[0.001, 0.05]0.0047489Fittedproportion[0.001, 0.05]0.012701Fittedproportion[0.001, 0.05]0.0045514Fittedproportion[0.00002, 0.005]0.00017377Fittedproportion[0.00001, 0.015]0.012362Fittedweek−1[40, 500]130.4989Fitteddimensionless[50, 1000]843.2311Fitteddimensionless[1.0085, 10.95]1.0247Fitteddimensionless[1.001, 20]2.201Fitteddimensionless[0.0001, 10.175]7.7161Fitteddimensionless[0.0025, 10.10]7.8704Fitted Model parameters, units, values, and source. Calibrated model parameters for VL transmission dynamics. The model calibration is performed using a multistart optimization algorithm to overcome the issue of local minima and obtain a globally optimal fit to the incidence data. This approach involves running multiple optimization searches from diverse initial parameter guesses and selecting the solution that minimizes the error between the model predictions and the observed case data. The resulting set of calibrated parameters provides a baseline representation of VL transmission dynamics in the region. Sensitivity analysis of parameters to the system To identify the key long-term drivers of model behavior, we performed a global sensitivity analysis on the ODE system (1) using Latin Hypercube Sampling (LHS)29,30. Partial Rank Correlation Coefficients (PRCCs) were calculated for the infected human compartment over a 350-month simulation. PRCC values outside the [-0.2, 0.2] threshold were considered to indicate moderate to strong influence, with positive values promoting infection and negative values suppressing it. To examine how parameter importance evolves over time, PRCCs were evaluated at three distinct epidemic stages: early (10 weeks), intermediate (250 weeks), and late (475 weeks) (Fig. 3). Sensitivity of parameters with respect to infected human, versus the time duration of the disease. At the early epidemic stage (approximately 3 months), three parameters exhibited strong positive correlations with infection burden: the vector-to-human transmission probability (, PRCC = 0.8263), the maximum biting rate (, PRCC = 0.8537), and the human-to-vector transmission probability (, PRCC = 0.3810). Among these, and demonstrated the highest sensitivity, indicating that vector-related parameters, specifically the rate at which susceptible humans acquire infection from infected sandflies and the biting frequency of vectors, dominate the early phase of infection dynamics. The disease-induced death rate in humans () exhibited a moderate negative correlation (PRCC = – 0.2877). The temperature effect modifying parameter () displayed a strong negative correlation (PRCC = – 0.7090). In contrast, reservoir-related parameters (, , ) and the remaining modification parameters (, , , ) showed negligible effects during this phase (Table 3). Table 3PRCC and significance levels for model parameters at early (10 weeks), intermediate (250 weeks), and late (475 weeks) epidemic stages. Highlighted rows indicate the statistically significant parameters.ParameterTime = 10 weeksTime = 250 weeksTime = 475 weeksPRCCp-valueSign.?PRCCp-valueSign.?PRCCp-valueSign.?0.75450.000000yes0.47830.000000yes0.35590.000000yes0.05950.188995no0.35690.000000yes0.44180.000000no0.16780.000192yes– 0.01520.738244no– 0.01700.707800no0.21160.000002yes0.67720.000000yes0.64900.000000yes– 0.27280.000000yes– 0.04510.319671no– 0.05750.204611no0.03070.497660no– 0.60580.000000yes– 0.75570.000000yes0.90220.000000yes0.88560.000000yes0.86240.000000yes0.06340.161825no0.65300.000000yes0.59460.000000yes– 0.78560.000000yes– 0.61070.000000yes– 0.56160.000000yes– 0.06190.171958no– 0.06270.166233no– 0.05680.210077no– 0.00830.854940no– 0.46570.000000yes– 0.41640.000000yes0.04070.369029no0.04680.301548no0.01810.689590noNote: PRCC values range from -1 to +1. Positive values indicate a positive correlation, while negative values indicate a negative correlation. Statistical significance is assessed at α = 0.05 PRCC and significance levels for model parameters at early (10 weeks), intermediate (250 weeks), and late (475 weeks) epidemic stages. Highlighted rows indicate the statistically significant parameters. Note: PRCC values range from -1 to +1. Positive values indicate a positive correlation, while negative values indicate a negative correlation. Statistical significance is assessed at α = 0.05 At the intermediate stage (60 months), the sensitivity pattern shifted notably. The vector-related parameters (PRCC = 0.6939) and (PRCC = 0.6621) remained highly significant positive drivers of infection burden, though their influence had slightly decreased compared to the early stage. Notably, the modification parameter () emerged as a strong positive driver (PRCC = 0.7369). The human-to-vector transmission probability () remained a significant but weaker positive driver (PRCC = 0.1206). The disease-induced death rate in humans () continued to show a significant negative correlation (PRCC = – 0.1125), maintaining its protective effect. The temperature-modifying parameter () exhibited a strong negative correlation (PRCC = – 0.6609), while the rainfall-modifying parameter () became significantly negative (PRCC = – 0.4724). This indicates that both temperature and rainfall modulation play important protective roles during the intermediate stage. The modification parameters and remained non-significant. Reservoir-related parameters continued to show negligible effects, further supporting the conclusion that zoonotic reservoirs have a minimal role in sustaining transmission in this setting (Fig. 4). Dynamics sensitivity analysis of model parameters influencing infected human population () at early (3 weeks), intermediate (150 weeks), and late (300 weeks) stages. PRCC values are color-coded: red (p<0.05, positive effect), blue (p<0.05, negative effect), and gray (p≥ 0.05, not significant). Significant parameters are marked with black triangles. Horizontal reference lines at -0.2 or 0.2 indicate the strong influence threshold. By the late epidemic stage (120 months), a distinct pattern emerged. The vector-to-human transmission probability (, PRCC = 0.6611), the maximum biting rate (, PRCC = 0.6606), and , PRCC = 0.7008 remained highly significant positive drivers, indicating that vector-related parameters continue to influence infection burden even at endemic equilibrium. The human-to-vector transmission probability (, PRCC = 0.1602) remained statistically significant, though its influence was comparatively weaker. The disease-related mortality to human continues to exert a protective effect even in the late stage. The temperature-modifying parameter (, PRCC = – 0.6195) and the rainfall-modifying parameter (, PRCC = – 0.5920) both exhibited strong negative correlations, highlighting the persistent protective role of environmental modulation. The modification parameters and remained non-significant throughout all stages. Reservoir-related parameters showed no significant effects across any time point, confirming the minimal role of zoonotic reservoirs in this specific epidemiological setting. The sensitivity analysis indicates several key insights for public health intervention strategies. First, vector control strategies targeting the vector-to-human transmission probability () and the maximum biting rate () would be effective across all epidemic stages, as these parameters consistently drive infection burden. Second, case management and prompt treatment of infectious individuals, which effectively reduce the human-to-vector transmission probability (), remain important but play a comparatively smaller role than vector-related parameters. Third, the parameters exhibiting consistent negative correlations ( and ) represent potential protective factors that modulate the effects of temperature and rainfall on vector mortality. These result suggest that interventions such as environmental management to increase vector mortality or reduce vector breeding success could amplify these protective effects. Finally, the reservoir-related parameters showed negligible influence across all stages, indicating that targeting zoonotic reservoirs would not be a cost-effective intervention strategy in this specific epidemiological setting. Instead, limited public health resources should be prioritized toward vector control, environmental management to reduce vector recruitment, and case management. Numerical simulation and discussion Prior to evaluating the impact of different interventions, we present the baseline simulation outcomes. The dynamics of the human population are illustrated in Fig. 5, while the reservoir and vector population trajectories are shown in Fig. 6, both generated using the baseline parameter and initial condition sets. We run the simulation for a final time of 1000 weeks (approximately 19 years) to capture the long-term effects of interventions across multiple transmission seasons and to observe the system’s behavior as it approaches endemic equilibrium. The dynamics of human population without impulsive controls. The time, t=0, corresponds to the first week of January 2010. The dynamics of sandfly and reservoir populations without impulsive controls. The time, t=0, corresponds to the first week of January 2010. Remark 1 The model assumes no recovery in the reservoir animal population, and hence those infected accumulate in the same cohort until they die. The effectiveness and coverage of different interventions determine the reduction rates for susceptible vectors, infected vectors, and infected reservoir hosts. These rates reflect the realistic limitations of implementing vector control strategies and achieving full coverage in the field. Considering these, we have formulated the number of susceptible sandflies, infected sandflies, and infected reservoirs reduced at each intervention period as follows: The number of susceptible vectors, both in their pre-adult and adult stages, can be significantly reduced following the implementation of vector control measures during each impulsive intervention period. Vector control through the destruction of breeding sites: one of the key strategies for controlling the sandfly population is the elimination of its breeding sites. By targeting and removing potential habitats where sandflies lay eggs and larvae develop, the overall vector population can be significantly reduced, thereby lowering the risk of disease transmission. Here, in this model, the reduction rate by these strategies is expressed as 1-p, where . With denoting the efficacy of the vector control measure, and representing its coverage. This coverage, the destruction of sandfly breeding sites, indicates the degree to which sandfly breeding sites, such as damp areas, organic waste sites, and cracks in walls, are successfully identified and eliminated during each intervention cycle. Since the destruction of sandfly breeding sites primarily targets the pre-adult stage sandflies, the efficacy of this type of vector control is assumed to be 100%. However, identifying and eliminating all possible breeding sites for sandflies is a challenging task, especially in remote and resource-limited settings. Therefore, the coverage of this intervention is assumed to reach a maximum of 55%. Consequently, the number of susceptible sandflies decreases at each intervention time by a factor of , where and . Vector control through the destruction of breeding sites: one of the key strategies for controlling the sandfly population is the elimination of its breeding sites. By targeting and removing potential habitats where sandflies lay eggs and larvae develop, the overall vector population can be significantly reduced, thereby lowering the risk of disease transmission. Here, in this model, the reduction rate by these strategies is expressed as 1-p, where . With denoting the efficacy of the vector control measure, and representing its coverage. This coverage, the destruction of sandfly breeding sites, indicates the degree to which sandfly breeding sites, such as damp areas, organic waste sites, and cracks in walls, are successfully identified and eliminated during each intervention cycle. Since the destruction of sandfly breeding sites primarily targets the pre-adult stage sandflies, the efficacy of this type of vector control is assumed to be 100%. However, identifying and eliminating all possible breeding sites for sandflies is a challenging task, especially in remote and resource-limited settings. Therefore, the coverage of this intervention is assumed to reach a maximum of 55%. Consequently, the number of susceptible sandflies decreases at each intervention time by a factor of , where and . Vector control strategy through the use of insecticide residual spraying: Insecticide residual spraying (IRS) is a crucial vector control approach aimed at reducing the transmission of VL by targeting sandfly populations. This method involves the application of long-lasting insecticides to the interior walls and surfaces of houses, animal shelters, and other resting sites of sandflies. By doing so, the IRS effectively kills vectors that come into contact with treated surfaces, thereby disrupting their life cycle and reducing their ability to transmit the disease. The maximum efficacy of IRS is assumed to be 90%, while the maximum coverage is considered to be 55%. This means that the reduction in the number of infected sandflies during each intervention period occurs at a rate , where and . Thus, at each intervention time, the number of infected sandflies will decrease by this rate. Vector control strategy through the use of insecticide residual spraying: Insecticide residual spraying (IRS) is a crucial vector control approach aimed at reducing the transmission of VL by targeting sandfly populations. This method involves the application of long-lasting insecticides to the interior walls and surfaces of houses, animal shelters, and other resting sites of sandflies. By doing so, the IRS effectively kills vectors that come into contact with treated surfaces, thereby disrupting their life cycle and reducing their ability to transmit the disease. The maximum efficacy of IRS is assumed to be 90%, while the maximum coverage is considered to be 55%. This means that the reduction in the number of infected sandflies during each intervention period occurs at a rate , where and . Thus, at each intervention time, the number of infected sandflies will decrease by this rate. Reservoir animals culling: The other intervention strategy to control the transmission of VL is removing or culling infected reservoirs, including infected dogs. However, the process of locating and ethically euthanizing all infected dogs within a specific region presents substantial logistical difficulties. It demands significant resources, careful planning, and coordinated efforts. Due to these challenges, the complete culling of all infected dogs is not a feasible or practical solution. Considering these challenges, completely culling all infected dogs is not a feasible approach. Therefore, we assume that the maximum achievable coverage of animal culling is 55%, with the intervention being fully effective in reducing disease transmission. Thus, the reduction in the number of infected dogs during each mass culling of reservoir animals occurs at a rate of , where and . Reservoir animals culling: The other intervention strategy to control the transmission of VL is removing or culling infected reservoirs, including infected dogs. However, the process of locating and ethically euthanizing all infected dogs within a specific region presents substantial logistical difficulties. It demands significant resources, careful planning, and coordinated efforts. Due to these challenges, the complete culling of all infected dogs is not a feasible or practical solution. Considering these challenges, completely culling all infected dogs is not a feasible approach. Therefore, we assume that the maximum achievable coverage of animal culling is 55%, with the intervention being fully effective in reducing disease transmission. Thus, the reduction in the number of infected dogs during each mass culling of reservoir animals occurs at a rate of , where and . Effect of the starting time of interventions in VL control The timing of interventions in VL transmission control is vital, as environmental factors such as temperature and rainfall significantly impact sandfly population dynamics, influencing their birth rates, survival, and biting behavior. These variations in the weather conditions drive the seasonal patterns of VL transmission2. The success of VL control measures is influenced by the timing of their implementation within the seasonal cycle. Different periods of the year, including the rainy season, dry season, and transitional phases with minimal rainfall, exhibit unique infection patterns. VL transmission fluctuates between peak and low infection periods, and the timing of these fluctuations plays a crucial role in determining the effectiveness of interventions. The first peak, representing the highest infection rate, occurs at t=9.5, which corresponds to early March, during the dry season with minimum rainfall, marking the first-highest infection period, when VL cases reach their maximum. The second peak follows at t=39.5, occurring in early October. On the other hand, the lowest infection rates are observed at two distinct points in the year. The first minimum occurs at t=28.5, aligning with mid of July, a period of heavy rainfall when VL cases are at their lowest. The second minimum follows at t=52, early January, which is the main dry season and marks the second-lowest infection phase. To assess the impact of initiating the fixed interventions at different times of the year, we simulate VL transmission under various control strategies. These interventions, including vector control measures and reservoir animal culling, are applied for two weeks for a total of one thousand weeks, either biannually (every six months) or annually. By implementing these measures at different infection peaks and troughs, we analyze their effectiveness in reducing VL transmission. The main objective here is to minimize the total number of human infections over the simulation period when interventions are initiated at different time windows, compared to the baseline scenario without any intervention. This study evaluates the effectiveness of various intervention strategies in reducing the total number of human infections over the entire intervention period. Specifically, the total number of infections averted under each intervention scenario is quantified and expressed as a percentage reduction relative to the baseline scenario, where no interventions are applied. By comparing these values, the study aims to identify the most effective strategies for minimizing disease transmission and controlling the spread of infection in the entire planning period. The results of these simulations are discussed in the following section. Since VL treatment follows a predefined medical protocol and requires a structured administration schedule, it cannot be effectively implemented as a short-term campaign-based intervention. Due to these constraints, we have not incorporated VL treatment as part of the control strategies analyzed in this study. Instead, our focus is on preventive measures, such as vector control and reservoir animal culling, which can be strategically timed and applied periodically to disrupt disease transmission. However, we acknowledge the critical role of VL treatment in managing individual cases and reducing disease severity. For our scenario analysis, we selected intervention start times relative to key epidemiological events: four weeks before the first peak (t=5.5), at the highest peak (t=9.5), four weeks after the first peak (t=13.5), and at the minimum point (t=28.5), as well as four weeks before (t=24.5) and four weeks after (t=32.5) this minimum. When intervention is employed once per year Our simulation results indicate that the timing for intervention initiation significantly impacts the effectiveness of control measures against VL. Based on our findings, the most effective strategy is to begin control measures at the minimum infection point, specifically at t=28.5 as indicated in Fig. 7e , which falls in mid of July (in Ethiopia). This optimal timing results in a 85.4% reduction in the total number of human infected cases by the end of the intervention period as referred in Fig. 8. Following this strategy, the second most effective intervention timing is to start 4 weeks before the maximum infection point at t=5.5. Initiating interventions at this point leads to a 69.21% reduction in VL cases. These moments present valuable alternative opportunities for intervention, underscoring the critical influence of well-timed control measures on disease dynamics. Simulation of the infected human population when both vector control and reservoir animals culling are applied once per year for two weeks, starting from different levels of infection. The time, t=0, corresponds to the first week of January 2010. Total infections saved at each intervention start time, with two-week interventions applied once per year. On the other hand, when interventions are initiated at the infection peak (t=9.5 weeks) or four weeks thereafter (t=13.5 weeks) as indicated in Fig. 7b and c , resulted in less pronounced case reductions, achieving 62.35% and 61.93%, respectively (see Fig. 8). While these timings do not achieve the same level of effectiveness as the optimal intervention window, they still contribute significantly to reducing the overall disease burden, underscoring that even delayed or early interventions offer meaningful epidemiological value. These findings provide compelling evidence that aligning intervention implementation with the minimum infection peak, which coincides with the rainy season, maximizes control effectiveness, resulting close to-elimination of disease transmission within the community. The results highlight the critical role of strategic intervention timing in controlling VL outbreaks. Specifically, deploying control strategies when infection numbers are at their minimum proves to be the most effective method for achieving sustained reductions in disease prevalence. When intervention is employed twice per year (every six months) The study further explores the impact of implementing biannual interventions, where control measures are applied every six months for two weeks. The findings reinforce the critical role of intervention timing in determining the overall effectiveness of disease control strategies, similar to what was observed in the case of annual interventions. A significant reduction in infection cases is observed when interventions commence at the minimum infection point, specifically at t=28.5, as in the case of applying intervention once per year. Initiating control measures at these point leads to the most substantial and total elimination with 96.6% overall decline in total infected cases in the intervention period (see Figs. 9e and 10). The second optimal intervention timing that can reduce the number of infected humans is initiating four weeks before the occurrence of the maximum peak, t=5.5, leading to a 70.8% reduction (Fig. 10). In contrast, when interventions are initiated at t=13.5, which corresponds to about one month after the maximum infection point leads to the least possible level of reduction, which is 54.9%. Simulation of the infected human population when both vector control and reservoir animals culling are applied every six months for two weeks, starting from different levels of infection. Total infections saved at each intervention start time, with two-week interventions applied twice per year. These findings underscore the critical importance of strategically timing interventions with seasonal variation of the disease dynamics to maximize the effectiveness of disease control measures. As demonstrated in Fig. 10, implementing control measures biannually yields significantly better outcomes in reducing disease prevalence compared to annual interventions. Specifically, when interventions are conducted twice per year, the number of total infected individuals decreases by an additional 11.2% compared to applying control measures only once per year. However, the success of these interventions is highly dependent on selecting the most effective starting points. The analysis suggests that initiating control measures at the minimum infection peak leads to the most substantial long-term benefits. This finding also highlights that periodic interventions initiated at less favorable points in the infection cycle may result in less effective disease control outcomes. These results provide valuable insights for optimizing intervention strategies. Based on our findings, we strongly advise policymakers and health authorities to align intervention initiation with the rainy season, the period of minimum infection prevalence, to maximize the effectiveness of transmission control efforts. By carefully scheduling vector control and reservoir animal culling within these optimal time frames, it is possible to enhance the overall effectiveness of disease management efforts. This approach not only ensures a more efficient allocation of resources but also contributes to a cost-effective and sustainable strategy for controlling VL. Ultimately, adopting such evidence-based timing strategies can lead to greater reductions in disease transmission and burden, improving public health outcomes in affected regions. In the subsequent sections, we examine the effectiveness of disease control strategies when implemented individually or in combination. Our analysis focuses on identifying the most impactful approach, particularly when interventions are initiated at the optimal timing, t=28.5, for a duration of two weeks, resulting in a more significant reduction in infection cases. Vector control The use of insecticide spraying and the management of vector breeding sites are among the most effective strategies for controlling vector populations in the fight against VL2. These vector control measures can be implemented in various ways to target sandflies at different stages of their life cycle. One key approach involves the systematic and thorough destruction of their breeding sites. This strategy focuses on eliminating the habitats where sandflies lay their eggs, thereby interrupting their life cycle and killing them at their pre-adult stages, including larvae and pupae. Another crucial strategy is indoor and outdoor insecticide residual spraying, which involves the application of insecticides to walls, surfaces, and other areas where sandflies may rest31. As noted by Abbasi44, vector control through residual spraying and community programs shows promise but is hindered by insecticide resistance, highlighting the need for innovative vector management strategies. This method ensures that sandflies coming into contact with treated areas are killed, further reducing the risk of disease transmission. By combining these methods targeting both the pre-adult and adult stages of the vector’s life cycle, vector control programs can effectively reduce the population of sandflies and play a significant role in controlling the spread of VL. Since the abundance of sandflies and the general activities (such as the biting rate, development cycle) of the vector depend on the weather conditions, applying vector control strategies based on these timings is very important to maximize impact and minimize the intervention costs. Considering these, in this article, vector control strategies are applied to the VL model at different specific times of the year, and the dynamics of the infected human population are determined. In this model, the population of susceptible vectors () and infected vectors () decrease at rates of 1-p and 1-q, respectively, during each intervention period. This expression is mathematically expressed in Eq. (13): where and , with and representing the efficacy of eliminating vector breeding sites and the efficacy of IRS, respectively. Similarly, and denote the coverage levels of these interventions. In this study, the effectiveness of vector breeding site destruction and IRS is assumed to be 100% and 90%, respectively. However, the maximum achievable coverage in practical scenarios is considered to be 55% for vector breeding site destruction and 55% for IRS, reflecting real-world constraints such as incomplete implementation, environmental factors, and operational limitations. Interventions are applied for two weeks or one week and compared with the baseline value, where the baseline value means the simulation of an infected human population without any controls. According to the results of the numerical simulation, implementing vector control interventions significantly reduces the number of infected humans. When this strategy alone is applied once per year for a duration of one week, the total number of infected humans decreases by 79.77% compared to the baseline scenario where no intervention is implemented. If the intervention duration is extended to two weeks, applied once per year, the reduction in the total number of infected humans becomes 83.29% (refer to Fig. 11). The intervention begins at t=28.5, which corresponds to six weeks later from the highest infection point. By implementing the intervention in the second week of February, which is six weeks after the highest peak, it is possible to curb further transmission, prevent secondary infections, and accelerate the decline in case numbers. This approach ensures that resources are allocated efficiently, targeting the critical phase of the epidemic when intervention can have the greatest impact in reducing disease prevalence. The simulation of infected human population when vector control only is applied once per year, starting at the time, t=28.5. Furthermore, when the intervention is carried out twice per year (every six months), each for one week, it results in a 91.83% reduction in the total number of infected humans. Similarly, increasing the intervention duration to two weeks, administered twice per year, leads to a reduction of 94.75% in the total infected human population within the intervention period (see Fig. 12). The intervention is scheduled to begin six weeks after the occurrence of the highest infection point, ensuring that control measures are implemented when the disease burden starts to decline. Following this initial intervention, subsequent rounds of the intervention are carried out at regular six-month intervals, maintaining a consistent and systematic approach to disease management. This structured timing aims to maximize the effectiveness of the intervention by targeting critical periods, thereby reducing the overall prevalence of the disease more efficiently. The simulation of infected human population when vector control only is applied every six months, starting at the time, t=28.5. These findings underscore the importance of both the frequency and duration of vector control measures in mitigating the spread of VL and highlight the significant impact of sustained and intensified intervention efforts in reducing disease burden. Reservoir animals culling The primary known reservoir for VL is dogs. Despite ethical concerns and challenges related to practicality, culling infected reservoirs has been employed as a preventive measure for VL in various countries. To evaluate the impact of reservoir animal culling as a sole intervention, we simulate the model while implementing only this control measure. When reservoir animal culling is implemented during each intervention period, the number of infected reservoir animals decreases at a rate of 1-m, where the reduction factor is given by . This reduction is mathematically represented in Eq. (14): Here, the effectiveness of the infected reservoir animal culling strategy, denoted by , is assumed to be 100%, meaning that all targeted animals within the intervention scope are successfully culled. Additionally, the maximum achievable coverage of the culling operation, represented by , is set at 55%. The simulation findings demonstrate that both the frequency and duration of reservoir animal culling substantially influence the reduction of visceral leishmaniasis (VL) infections. When infected reservoir animals are culled once annually, the effectiveness varies with intervention duration. Specifically, a one-week culling campaign reduces human infection cases by 67.46% relative to the baseline scenario. However, extending the culling period to two weeks diminishes the reduction to 66.16% (Fig. 13). The simulation of infected human population when reservoir animals culling only is applied every once per year, starting at the time, t=28.5. Furthermore, increasing the frequency of culling from annual to biannual schedules reduces its overall effectiveness in controlling VL transmission. Implementing one-week culling twice per year results in a 59.72% reduction in infected human cases, whereas two-week biannual culling lowers infections by only 56.9% (Fig. 14). The simulation of infected human population when reservoir animals culling only is applied every six months, starting at time, t=28.5. These results highlight a critical and counterintuitive insight: longer culling durations and higher culling frequencies do not necessarily translate into greater disease control. This finding suggests that investing additional resources in prolonged or more frequent reservoir culling campaigns may not be necessary as an effective VL control strategy. Health authorities should therefore prioritize short-duration, strategically timed culling interventions over extended or intensified campaigns. When resources are limited, alternative strategies, particularly vector control or combined interventions, offer more reliable and cost-effective approaches to reducing VL transmission. We acknowledge that reservoir animal culling raises important ethical and social considerations. Ethically, culling must be justified by clear public health benefits and conducted humanely, adhering to animal welfare standards. Socially, such interventions may face community resistance, particularly where animals hold economic, cultural, or companion value. Transparent communication, community engagement, and culturally sensitive implementation are essential for program acceptability. Where feasible, non-lethal alternatives (e.g., vaccination or habitat management) should be considered. These considerations, while beyond the scope of our modeling study, warrant attention in policy formulation. Combining Reservoir animals culling together with vector control In this section, we integrate reservoir animal culling alongside vector control to evaluate their combined effectiveness in reducing disease prevalence. To capture this impact, we incorporate mathematical expressions into the model to quantify the reduction in infected sandflies and reservoir animals resulting from these intervention efforts as outlined below: where , and are as defined before in Eq. (13) and Eq. (14). The simulation results emphasize the significant impact of integrating infected reservoir culling with vector control strategies in reducing VL transmission. The effectiveness of these interventions is strongly influenced by both their duration and frequency of application. When implemented annually, the combined intervention leads to a progressive reduction in the number of infected humans. Specifically, applying control measures for one and two weeks per year results in a decline of 84.43% and 85.4%, respectively (see Fig. 15). The simulation of infected human population when both vector control and reservoir culling are applied once per year, starting at the time, t=28.5. A more frequent intervention schedule, where vector control and reservoir culling are conducted biannually (every six months), yields even more substantial reductions. Under this approach, applying interventions for one and two weeks per cycle results in a 95.11% and 96.60% reduction in the total number of infected humans, respectively (see Fig. 16). Notably, when control measures are implemented twice a year for two weeks per cycle, the reduction in infections is remarkably high. The simulation of infected human population when both vector control and reservoir culling are applied every six months or twice per year, starting at the time, t=28.5. These findings highlight the crucial role of integrating multiple intervention strategies and optimizing their frequency and duration to achieve effective disease control. The results suggest that more frequent and extended applications of vector control and reservoir culling significantly enhance disease reduction, underscoring the need for a well-structured, sustained intervention plan to minimize VL transmission effectively. The impact of frequency of interventions on disease control effectiveness The frequency of interventions is a key determinant of disease control success, yet optimal scheduling remains poorly understood due to complex interactions between intervention timing, pathogen transmission dynamics, and ecological responses of vector and reservoir populations. To address this, we investigated the effect of increasing intervention frequency from one to two weeks and from once per year to twice per year across different control strategies. As shown in Fig. 17, advancing intervention frequency from one to two weeks resulted in a marked reduction in total infections saved, particularly for vector-based strategies where frequent suppression prevents rapid vector population resurgence. The comparison to show the effect of intervention frequency. Increasing intervention frequency from once to twice per year substantially enhanced infection reduction across all strategies. Combined controls achieved the highest effectiveness, saving 95.11–96.60% of infections when applied twice yearly, compared to 84.43–85.40% with annual application-an absolute improvement of approximately 10–11 percentage points. Vector control alone also responded favorably to increased frequency, with biannual applications (91.83–94.75% saved) outperforming annual applications (79.77–83.29% saved) by roughly 12%. In contrast, reservoir culling alone showed the smallest gain from increased frequency (approximately 8–9%), and notably, two-week interventions performed slightly worse than one-week interventions in the twice-yearly scenario (56.90% compared with 59.72%), suggesting potential compensatory ecological dynamics such as enhanced reproduction or immigration following prolonged culling. Regarding intervention duration, vector control consistently benefited from extended two-week applications, achieving 2.9–3.5% additional savings compared to one-week interventions across both frequencies. Combined controls showed modest improvements (approximately 1.5%) with longer duration. However, reservoir culling alone exhibited diminishing or negative returns from extended duration, particularly under twice-yearly application, indicating that shorter, more frequent culling may be preferable to avoid counterproductive ecological responses. The substantial effectiveness gap between vector control and reservoir culling alone (approximately 38% as indicated Fig. 17 panel b) suggests that vector suppression is the dominant driver of infection reduction, likely due to the central role of sand flies in transmission. Twice-yearly combined interventions approaching 96.60% effectiveness indicate near-elimination potential in moderate-to-high burden settings. These findings support prioritizing intervention frequency over duration when resources are constrained, avoiding prolonged reservoir culling as a standalone strategy, and emphasizing combined vector and reservoir control for optimal outcomes. The impact of interventions with different level of coverage Coverage is a critical factor in disease prevention, as it determines the extent to which protective measures and interventions reach the target population, thereby effectively disrupting the transmission cycle. Achieving high coverage is essential to ensuring widespread protection, interrupting disease transmission, safeguarding vulnerable groups, and ultimately reducing the overall burden of the disease. In this study, varying levels of coverage for VL control strategies were carefully incorporated into the impulsive mathematical model to examine their impact on the progression and dynamics of the infected human population over time. Mathematically, these interventions are expressed as: with , and as indicated in Eq. (15). Here, , and the coverage rates for destroying vector breeding sites, applying indoor residual spraying (IRS), and culling infected reservoir animals (such as dogs), respectively. To explore the sensitivity of these interventions, we increased the coverage levels of each control measure by 10%, 20%, and 30%. The simulation results, illustrated in Fig. 18, reveal that higher intervention coverage significantly reduces the prevalence of infection in the human population. Specifically, when these interventions (destruction of breeding sites, residual spraying, and infected dog removal) are applied twice a year for two-week durations at an additional 10% coverage level, the total number of infected humans decreases by 97.9%. Increasing the intervention coverage by just 20% from the baseline(or 75% coverage) results in a remarkable 98.55% reduction in the total number of infections. Moreover, applying all three interventions at a minimum of 85% coverage leads to a 98.89% decline in human infection cases (see Fig. 18). These findings underscore the critical importance of achieving and maintaining high levels of intervention coverage in the fight against VL transmission. While partial interventions can yield moderate reductions, sustained efforts at higher coverage levels lead to far more significant public health impacts. The results of this study reinforce that scaling up intervention coverage is key to effectively reducing disease burden and moving toward the long-term control and possible elimination of visceral leishmaniasis. The simulation of infected human population when both vector control and reservoir animal culling are applied with an additional 10%, 20%, and 30% intervention coverage to the baseline every six months for two weeks. These findings highlight the crucial role of increasing intervention coverage in controlling VL transmission. Furthermore, it suggest that while expanding coverage alone contributes to a significant decline in infection prevalence, more frequent implementation of these interventions could further enhance their effectiveness. This emphasizes the importance of optimizing both the scale and frequency of control measures to achieve a more substantial and sustained reduction in VL cases. Conclusion and recommendation This study introduces a VL model that incorporates impulsive perturbations, with some model parameters dynamically influenced by temperature and rainfall. By focusing on the principal preventive measures: vector control and animal culling, the model integrates targeted interventions such as indoor residual spraying with insecticides, the destruction of sandfly breeding sites, and the removal of infected dogs. These strategies aim to significantly reduce vector and reservoir populations, thereby decreasing the total prevalence of the disease substantially. The VL transmission dynamics were analyzed within the framework of hybrid dynamical systems, which capture continuous processes interrupted by discrete, impactful events. These interruptions, such as the immediate reduction of vector populations through vector control or the elimination of infected reservoirs via dog culling, provide a realistic representation of intervention mechanisms. A rigorous qualitative analysis of the model revealed that the disease-free periodic solution is locally asymptotically stable when the basic reproduction number , ensuring disease eradication under effective control. Conversely, the disease persists when , underscoring the critical threshold for intervention success. Model Calibration and Parameter Estimation were performed to estimate the key model parameters governing visceral leishmaniasis transmission. The calibration process successfully estimated these epidemiological parameters for Abdurafi, Ethiopia, using a multistart optimization algorithm to ensure a globally optimal fit to the 66-month incidence dataset. This close correspondence between model predictions and empirical data validates the model’s ability to replicate real-world transmission dynamics, thereby establishing a credible foundation for evaluating intervention strategies under seasonal climatic forcing. The global sensitivity analysis revealed that vector-related parameters, particularly the vector-to-human transmission probability () and the maximum biting rate (), consistently exerted the strongest positive influence on infection burden across all epidemic stages (early, intermediate, and late), with PRCC values ranging from 0.66 to 0.90. In contrast, reservoir-related parameters (, , and ) showed negligible effects throughout the entire simulation period, indicating that zoonotic reservoirs play a minimal role in sustaining transmission in this specific epidemiological setting. The temperature and rainfall-modifying parameters ( and ) exhibited consistently strong negative correlations (PRCC ranging from – 0.79 to – 0.42), highlighting the protective role of climatic factors that increase vector mortality or reduce breeding success. These findings carry critical public health implications: vector control strategies targeting biting rates reduction should be prioritized as the most effective intervention across all epidemic phases. Environmental management to enhance vector mortality represents a promising supplementary approach. Crucially, targeting zoonotic reservoirs is unlikely to be cost-effective, and limited public health resources should instead be directed toward vector control, environmental management, and prompt case detection and treatment. Numerical simulations emphasize the critical role of strategically timed and integrated interventions in controlling VL transmission. Environmental factors such as temperature and rainfall significantly influence sandfly population dynamics, regulating their birth rates, survival, and biting behavior, which in turn drive the seasonal fluctuations of VL transmission. Our findings highlight that the effectiveness of interventions varies depending on whether they are implemented at the highest peak or the lowest point of infection, as well as on the frequency, duration, and coverage of the control measures. The study also highlights the critical importance of intervention timing in effectively controlling and reducing the transmission of VL.Strategically initiating control measures at the point of minimum infection yields the most substantial reduction in total VL cases. Specifically, a two-week intervention launched at this minimum point (t=28.5 weeks), which corresponds to mid-July in Ethiopia during the rainy season, achieves a 85.4% reduction in human infections. In contrast, starting the intervention four weeks before the maximum infection peak (t=5.5 weeks) results in a 69.2% reduction when applied once per year. While interventions initiated at the peak or several weeks before and after this period also produce considerable case reductions, they remain comparatively less effective than those timed at the optimal points. These findings indicate that aligning interventions with seasonal disease patterns, particularly during heavy rainy seasons, maximizes the impact of control efforts. Furthermore, increasing the intervention frequency to twice per year for every six months while maintaining a two-week duration significantly enhances VL reduction, leading to a 96.6% decrease in infected cases when initiated from the same infection point, t=28.5. Beyond optimizing timing, this study underscores the necessity of combining multiple control strategies for more effective VL suppression. When vector control alone is applied once per year for two weeks, the infected population decreases by 83.29%. Doubling the frequency of intervention to every six months further enhances disease reduction, achieving a 94.75% decline. Similarly, reservoir animal culling alone reduces cases by 66.16% when applied annually, whereas and increasing the frequency to twice per year for a two-week periods doesn’t give a better reduction. While vector control is a widely used approach, its effectiveness is significantly improved when combined with reservoir animal culling. Our findings indicate that implementing both strategies once per year for one week reduces VL cases by 84.43%, and extending the intervention duration to two weeks leads to a 85.40% reduction. Applying the same interventions biannually further amplifies disease control, yielding a 95.11% reduction for one-week interventions and a 96.60% decline when extended to two weeks. Moreover, applying a maximum of 85% intervention coverage reduces the human cases by 98.89% and increasing the coverage correspondingly results in a corresponding reduction in infected cases. These findings highlight the crucial role of intervention coverage, duration, and frequency in maximizing disease control outcomes. Overall, this study underscores the necessity of integrating multiple intervention strategies rather than relying on a single approach. Based on our findings, we strongly recommend commencing VL control measures during the rainy season when infection become minimum, which occurs in the second week of July in Ethiopia. In contrast, interventions launched at times other than t=28.5 are less effective in achieving sustained long-term reductions. Targeting these critical transmission periods with well-coordinated two-week interventions ensures a more substantial and sustained reduction in cases. Additionally, achieving optimal intervention coverage, particularly through a combination of vector control, insecticide spraying, and reservoir animal culling, significantly enhances disease suppression efforts. This integrated approach not only accelerates the decline in infections but also increases the likelihood of long-term disease elimination. To maximize impact and ensure efficient resource allocation, policymakers and public health officials need to prioritize these strategic intervention periods. By adopting a targeted, high-coverage approach during the most effective seasonal windows, it is possible to move toward the complete elimination of VL in endemic regions. The findings offer valuable insights for public health officials and policymakers, enabling them to design time-sensitive, resource-efficient intervention strategies that align with the dynamics of VL transmission. Ultimately, this study contributes to the broader goal of sustainable and cost-effective disease management, ensuring that interventions are both impactful and feasible, particularly in resource-limited settings. This study has four main limitations. First, it does not incorporate a cost-effectiveness analysis of the different intervention strategies. While our findings demonstrate the comparative epidemiological effectiveness of different control approaches, the absence of economic evaluation limits practical guidance for resource-constrained health systems where budget considerations are paramount. Second, the model was validated only using data from one of endemic countries, Ethiopia, and has not been tested against time-series data from other VL-endemic countries. Therefore, the generalizability of our results to different epidemiological settings, characterized by distinct transmission dynamics, vector species, reservoir hosts, and climatic patterns, requires further validation. Future research should address these gaps by integrating economic analyses and cross-country validations to support evidence-based policy decisions.Third, our model assume that reservoir animals, once infected, do not recover and remain infectious until death. This assumption may not hold for all reservoir species, some of which may clear infection or develop immunity. Fourth, our model assumes homogeneous mixing between human, vector, and reservoir populations. This assumption oversimplifies the well-documented spatial and ecological heterogeneity of VL transmission, where infection risk is often clustered in specific households, villages, or ecological niches. Future spatially explicit models would be needed to capture such fine-scale heterogeneity. Future models incorporating reservoir recovery dynamics could provide a more nuanced assessment of transmission persistence and control effectiveness. Acknowledgements We are thankful to the editors and the anonymous reviewers for the valuable suggestions to improve this paper. Appendix A: Mathematical interpretation of model function Sandflies thrive under warm climatic conditions, with optimal development and survival observed at temperatures ranging from 18–28° C24 and moderate monthly rainfall levels up to 120 mm32. Favorable weather enhances key sandfly activities, including host-seeking, biting, blood-feeding, and rates of development and reproduction45. Conversely, prolonged exposure to temperatures below 15° C or above 32° C adversely affects sandfly survival, leading to increased mortality24. Furthermore, because sandfly larvae develop in moist soil or organic matter, heavy rainfall can destroy breeding sites and larval habitats, thereby reducing the sandfly population. Excessive precipitation may also directly wash away adult sandflies or impede their movement. The parameter , for the bitting rate of sandfly, is defined in this study as where W(T) accounts for the effect of temperature on a weekly average bitting rate of sandflies, and is the maximum biting rate of the sandfly per week. Similarly, the parameters and are defined as Here, H(T) and K(Υ ) describe the effect of temperature and rainfall, respectively, on the weekly vector mortality rate, while M(T) accounts for the temperature-dependent weekly recruitment rate of sandflies. The modification parameters , , and ensure these rates remain between 0 and 1, controlling their activation according to favorable temperature and rainfall conditions. Unlike mosquitoes, sandflies lack an aquatic life stage and complete their entire life cycle on land. Consequently, their development and survival are strongly governed by ambient temperature. Temperature critically regulates key physiological processes, including developmental rates, reproductive cycles, and mortality34. It directly influences the duration of egg hatching, larval maturation, and adult emergence, thereby playing a pivotal role in shaping sandfly population dynamics and seasonal activity patterns33. The temperature and rainfall dependent component functions are defined as with (corresponding to a one-year period in weeks), and are weekly average temperature and rainfall respectively, are the amplitude of temperature, , , , and are the amplitude of rainfall, and are temperature and rainfall values that are known to be most favorable for sandflies respectively, and are the phase shift for temperature as well as , , and are the phase shift for rainfall, , , and are the frequencies for rainfall, σ is the standard deviation from the mean temperature and ν is the standard deviation from the mean rainfall35 and are modification parameters. According to24, sandfly mortality increases when temperatures exceed or drop below . Thus, within the range of to , the climate is favorable for sandfly reproduction and survival, whereas temperatures outside this interval accelerate their death rate. Based on this observation, we formulated the modification parameter as a function of temperature, as presented in Eq. (17) and illustrated schematically in Fig. 6(a). Furthermore heavy rainfall can flood the sites where sandfly larvae develop, the rainy season and colder temperatures are unfavorable for sandfly populations. Intense rainfall hampers sandfly activities, including flying, biting, and developmental processes. Beyond a certain rainfall value, the rain suffocates and kills the larvae of sandflies. Considering these into account, we define a piecewise modification parameter in Eq. (3) and diagrammatically as indicated in Fig. 6(b). Furthermore, heavy rainfall can flood the larval development sites of sandflies, making both the rainy season and colder temperatures unfavorable for sandfly populations. Intense precipitation hinders critical activities such as flight, biting, and development. Beyond a certain threshold, rainfall may also suffocate and kill sandfly larvae. Taking these factors into account, we introduce a piecewise modification parameter, , defined in Eq. (3) and illustrated diagrammatically in Figs. 6(b), 19. The plot of the modification parameters and . Substituting the estimated parameter values and simplifying gives the temperature and rainfall dependent model parameters as follows. Author contributions T.D. Awoke: Formulation and analysis, investigation, methodology, validation, writing original draft, writing review and editing. S. M. Kassa: Conceptualization, methodology, project administration, writing review, editing and supervision. G. M. Tsidu: Funding acquisition, validation, project administration, writing review, editing and supervision. Funding This research was funded by O.R. Tambo Africa Research Chairs Initiative as supported by the Botswana International University of Science and Technology, the Ministry of Tertiary Education, Science, and Technology; the National Research Foundation of South Africa (NRF); the Department of Science and Innovation of South Africa (DSI); the International Development Research Centre of Canada (IDRC); and the Oliver and Adelaide Tambo Foundation (OATF), with grant number: (UID)136696. Data availability The data used to support the findings of the study are available within the article. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data used to support the findings of the study are available within the article.