Effects of Temperature Dependence in Mosquito Mortality on Simulated Chikungunya Virus Transmission
Florida Medical Entomology Lab, University of Florida-IFAS, Vero Beach, FL 32962, USA
Viruses 2025, 17(11), 1486; https://doi.org/10.3390/v17111486
Submission received: 31 July 2025
/
Revised: 30 September 2025
/
Accepted: 21 October 2025
/
Published: 8 November 2025
(This article belongs to the Section Human Virology and Viral Diseases)
Abstract
A compartmental, deterministic model was used to explore the effects of temperature dependency in mosquito mortality on the likelihood of epidemics and the size of outbreaks of Chikungunya virus under Florida temperature conditions. Two known vectors, Aedes albopictus and Ae. aegypti, were included, with similar structure but allowing mortality and abundance parameters to vary between them. The mortality relationship with temperature had a central optimal survival region, with increasing mortality outside these regions. The central temperature and the annual mean temperature were most influential in the likelihood of an epidemic, although the variance explained was low. The central temperature, annual mean temperature and day of virus infection influenced the size of the outbreaks. Regression models including two-way interactions explained more of the variance in outcomes than the main effects models, but there was still substantial variance left unexplained. Given the model structure, higher order interactions would be required to explain most of the variance.
1. Introduction
Vector mortality has been identified as a critical component of vector-borne pathogen transmission in many models (e.g., [1,2,3,4]; reviews in [5,6,7]). Increased survival (lower mortality) allows more individuals to survive the extrinsic incubation period and have a higher expectation of infectious life, increasing vectorial capacity and R0. For Ae. aegypti and Ae. albopictus, as well as other mosquitoes, mortality has been shown to vary with temperature in numerous studies. Meta-analysis and pooled analysis [6,7] have shown high variation between studies, suggesting that factors other than temperature influence mortality. These studies show variation in the relationship between mortality and temperature. While there is often a range of temperatures optimal for mosquito survival (lower mortality), the specific temperatures and the width of the optimal range vary between the data sets and analyses performed [6,7,8]. Many factors likely contribute to this variation (e.g., environment, genetics, experimental details, analysis) but the specific relationships are not known.
Chikungunya virus (CHIKV) is a mosquito-borne arbovirus, most commonly transmitted by Ae. aegypti. It can also be transmitted by Ae. albopictus, and some lineages have evolved to better use Ae. albopictus as a vector [9]. While these two species are similar in their ecology, there are differences, notably in host use, with Ae. aegypti having a stronger preference for humans, but also in habitat preference or tolerance [10,11,12,13] and in the mortality–temperature relationship [3,6,7]. The two species can coexist, and in those cases both species can contribute to CHIKV transmission. In Florida, the two species can be found in the same containers and overlapping habitat [14] but can also show an abundance gradient from coastal (more Ae. aegypti) to inland (more Ae. albopictus) areas [15]. The coastal–inland gradient generally overlaps with the urban–rural gradient, as the coast is more urban. Further investigation [16] showed that the factors explaining the presence of each species overlapped at the site level but showed differences in remotely sensed variables. Lwande et al. [17] reviewed the global range and characteristics of the two species, demonstrating that their ranges overlap substantially and there are areas of sympatry. In endemic regions, there are CHIKV cases each year and periodic outbreaks, and the virus can be introduced by travelers to new areas, resulting in outbreaks [18,19]. The relative roles of the two mosquito species varies in space and time. In the US, there was an outbreak in 2014–2016 involving multiple states and territories (e.g., [20,21], including Florida [20]. There are introductions each year from returning travelers [18], presenting an ongoing risk of introduction and epidemics. Improving our understanding of the dynamics of this virus and the factors influencing the likelihood of outbreaks following an introduction is important.
Previously, we developed an arbovirus–mosquito transmission model which has been adapted for different viruses (St. Louis Encephalitis virus/West Nile virus [22,23]; CHIKV [3]; Dengue virus [4]). The CHIKV model included the annual temperature cycle as an input variable and focused on Florida, where the amplitude and mean of the annual cycle are correlated [3]. Using this model, we showed that differences between the two mosquito species and the mean annual temperature affected transmission dynamics, and that aspects of the mortality–temperature relationship altered the importance of other factors.
Our CHIKV model uses a temperature–mortality relationship based on the studies above for both mosquito species. This relationship includes an optimal region where mortality is at its lowest value, increasing to high levels outside the optimal range. In the previous analysis [3] the width of the optimal range was set fairly narrow (2 °C) and only three values of the central point of the optimal survival range were used. This minimal variation in the optimal region, however, showed a strong impact on model behavior and the relative importance of the other parameters in the model. Therefore, we conducted a second analysis of the model, focusing on the mortality and temperature parameters, to assess how these parameters would affect the likelihood and size of an epidemic when varied over a larger parameter space.
2. Materials and Methods
The Aedes-CHIKV model reported in Lord et al. [3] was used to follow up on results from that study. Specifically, this analysis focuses on the mortality parameters and interactions with temperature and seasonality. As in the previous model, Latin Hypercube sampling (e.g., [3,4,22,23,24]) was used to develop parameter sets. Parameters, descriptions, ranges and distributions for sampling are given in Table 1; the full parameter set used to run the simulations is given in Table S1. Here, most parameters were fixed at values permissive for epidemics based on the previous analysis, and 10 parameters related to mortality and seasonality were varied. For each parameter varied, a distribution is assigned along with minimum, maximum and most probable value (for triangular distributions). Distributions are sampled on the probability axis, generating more values around the most probable values for triangular distributions, then randomly resorted into parameter sets to run the model. Here, we used 250 unique and independent parameter sets.
Full details of the model, mosquito population dynamics, temperature curves and mortality function structure, including parameter estimation, can be found in Lord et al. [3] (see Figures 1 and 2 therein). A brief outline of the model structure and key details is provided here for reference. This is a compartmental, deterministic model with an SIR structure for hosts (humans) and an SEI structure for vectors (two species of mosquito, Ae. albopictus and Ae. aegypti). An exposed class was not included for humans, as our focus was how different vector species influence transmission cycles and there is no evidence that the vector species affects the duration of the incubation period in humans. Additionally, the incubation period is shorter in CHIKV than in other arboviruses [25], so it would have less impact on the dynamics. A daily temperature is used as a driver for temperature-based virus development and mortality in mosquitoes. Daily temperatures are calculated based on the mean of the annual temperature curve, Tmean, using a relationship derived from Florida temperature data. In Florida, the amplitude of the annual temperature curve is correlated with the mean, and the mean and amplitude are used in a sinusoid function to calculate daily temperatures. Virus is introduced into the system with one infectious human on day tcrit, varied over the entire year in both analyses.
Immature mosquitoes are not explicitly modeled, and adult patterns are functions of mosquito recruitment and mortality. Recruitment is based on general rainfall patterns for Florida and are constructed as a baseline level of recruitment (pulses at iv intervals year-round) and two seasonal peaks (spring—early center; mid-late summer) centered on days q1 and q2. Model structure allows for variation in the timing and relative size of these recruitment phases, but recruitment structure was not varied in the previous or current analysis; variation in recruitment is planned for future work. The recruitment pattern used here is based on observed rainfall patterns and mosquito abundance in south Florida. Total recruitment (Rtot) is varied separately for the two species, to allow for abundance to impact transmission. Following Lord et al. [3], the biting rate aj is the inverse of the days between blood meals on humans, αj. Aedes albopictus can be more opportunistic than Ae. aegypti, more readily feeding on non-human hosts (e.g., [26,27,28]. Aedes aegypti, however, may feed more frequently [29,30]. In both cases, high variation between populations has been observed. To reflect these differences, previously the most probable value of αj was set lower for Ae. aegypti (3 vs. 5, Table 1) and the parameters varied in the analysis. Here, we used these most probable values as fixed parameter values.
Mosquito mortality is temperature dependent, with a central temperature region of optimal survival (low mortality), high mortality at low and high temperatures and linear phases between these regions (see Figure 2 in [3]). The center point of the optimal survival region (Tempc) and the width to either side (W) specify the location and size of the optimal survival region and are the same for the two species. The minimum mortality in the central region and the slope of the increase outside this region (μmin, μsl) further determine the mortality rate at each temperature; these parameters are sampled separately for each species, although the ranges are set the same in this analysis. Previously, three values were used for Tempc and W was fixed at 2 days; here Tempc was varied between 10 and 20 °C and W from 2 to 5 days, to generate a wide range of possible patterns reflecting the variation observed in studies of mortality. This generated mortality–temperature patterns that likely extend beyond the current range of these species, to consider more extreme conditions and explore possible consequences on transmission. Transmission parameters (Table 1 and Table S1) were fixed at permissive values from the previous analysis, to allow for sufficient epidemics in the model set for analysis. There is high variance in mosquito competence for CHIKV, between species, CHIKV strains and with CHIKV mutations that increase the competence of Ae. albopictus (e.g., [31,32,33]). Lord et al. [3] used broad ranges for the two transmission parameters, with the ranges and most probable value set the same but the distribution sampled independently. These parameters did influence outcome variables, although Ae. aegypti parameters were more influential overall. Here, these parameters were fixed at values that were previously permissive for outbreaks, to focus attention on the mortality parameters while generating enough epidemics for analysis.
Virus development in the mosquito is temperature dependent, and this combined with seasonal reproduction and temperature-dependent mosquito mortality creates seasonality in transmission between mosquitoes and humans.
The model was updated and simulations run in Matlab (version 2019a); analysis was performed in Matlab (version 2021a). The model code for the ODE model is given in the Supplementary Data for Lord et al. [3]. The parameter sets are given in Table S1.
Output variables and statistical analysis: Output variables were chosen to focus on the likelihood of epidemics and the size of epidemics, as in the previous analysis. For human infections, the peak number of infected humans during the simulation (MaxHi) was assessed for each simulation run. If MaxHi > 2 the run was scored as epidemic and the variable epi set to 1; if not, epi = 0. This characterized each run as epidemic (epi = 1) or not (epi = 0), dividing the 250 simulation runs into the two groups. Epi and MaxHi (epi = 1 (epidemic) runs only) were analyzed for their relationship with input parameters.
Using the Latin Hypercube sampling structure, the parameters are sampled independently and should be orthogonal, allowing for regression analysis on the output variables. In some cases, however, parameters for Ae. aegypti and Ae. albopictus were sampled independently but from the same distributions (Table 1). This can occasionally create correlations between the two realized sets of parameter values. Correlation analysis between the 10 parameter sets showed one strong correlation between two parameters sampled from the same distributions, Rtot,alb and Rtot,aeg (Table S2). Initially, the analyses were performed with all 10 sampled parameters; however, the matrices for these analyses were ill-conditioned and the results are not reliable. Based on the correlation analysis, we tested dropping one of the correlated parameters from the analysis (Rtot,aeg); these analyses were better-conditioned and reliable and are reported here.
Logistic regression was used for epi (Matlab function fitglm with distribution “binomial), while linear regression was used for maxHi (Matlab function fitglm with distribution “normal”). All varied input parameters were included in the regression models, except for dropping the correlated parameter (Rtot,aeg). In the model situation, all the parameters will have some effect, unlike a field situation where it may be unknown if a measured variable actually influences the outcome variable. Because of this, rather than focusing on p-values that meet predetermined thresholds (e.g., α = 0.05 is significant), we use the p-values to rank parameters in order of importance. Parameters and associated p-values were sorted from smallest to largest p-value, then the parameters with the smallest p-values were characterized as the most influential on the output variable. For both output variables, we first did a main effects model, to assess which parameters had the most effect on outcomes directly. We then included two-factor interactions in the regression models to assess the effect on the variance explained and the influence of each input parameter. For the main effects model (nine main effects + intercept) we focus on the three parameters with the lowest p value; for interaction models we expanded this to the top five.
3. Results
3.1. Model Behavior
As in the previous sensitivity analysis of this model, we observed a range of model behaviors (Table S3). Many of the simulations showed epidemic behavior (206/250 simulations), in part due to the permissive parameters used. The most common behavior (number of runs, n = 192) was a single epidemic including most hosts (Figure 1, left panels), with some variation in the time between virus introduction and the peak of the epidemic and in the length of the tail. A few simulations (n = 6) were epidemic (maxHi > 2) but maxHi was very low, indicating the virus died out quickly. Some simulations (n = 14) showed bimodal outbreaks with low virus activity in between (Figure 1, right panels), and the relative sizes of the two outbreaks varied. Frequently, all hosts became infected during the simulation.
3.2. Likelihood of Epidemics
In the main effects model, the central temperature (Tempc) was the most influential variable in the likelihood of an epidemic (epi) (Table 2, Figure 2), followed by the slope of the mortality function in Ae. aegypti (μsl,aeg, Table 2). The day of virus introduction was the third parameter in the top three. However, the main effects model only explained 8% of the variance (adjusted R2 (adj. R2) = 0.08, Table 2). The full two-way interactions and main effects model explained 45% of the variation. In this model, the parameter/interactions that were most influential (top five, due to the number of pairwise interactions) included Tmean, Tempc, W, tcrit and the slope of the mortality function for Ae. albopictus (μsl,alb) (Table 3 and Table S4A). The interaction with the lowest p-value was Tempc * Tmean (Table 3, Figure 2), although the effect was not strong.
3.3. Size of Epidemics
For the size of the epidemic (MaxHi), Tempc was again the most influential variable in the main effects model (Table 2, Figure 3), followed by the maximum recruitment for Ae. albopictus (ρmax,alb). The width of the optimal survival region (W) was the third parameter in terms of influence on the outcome variable. Tempc effects are not easily visualized, but there is a trend for increasing MaxHi with increasing Tempc. As with epi, the effects are not strong and the main effects model only explained 16% of the variation (adj. R2 = 0.16). In the interactions model, the most influential variables were interactions between Tmean and tcrit (Figure 4, Table 3 and Table S4B) and Tempc and μsl,alb. High values of tcrit with low values of Tmean generated a few epidemics with low maximum numbers of hosts infected, although again the signal was not strong. When adjusted for the number of variables, the interaction model explained less of the variance in the data (adj. R2 = 0.131) than the main effects model, although the unadjusted R2 was higher in the interaction model (R2 = 0.32). Multiple interactions in the top five included both Tempc and Tmean, indicating that these variables are influential in many ways.
4. Discussion
There was substantial variation unexplained by the models, indicating that higher order interactions are contributing to the model dynamics. Notably, main effects models explained only minor amounts of variation in the outcome variables. Given the structure of the model, with interacting parameters in most of the differential equations, this is not surprising. For the likelihood of epidemics, the inclusion of two-way interactions indicated that outbreaks appear slightly less likely when Tempc is low and Tmean is high. There is a region of Tmean around 22 °C where epidemics appear more likely (Figure 2, region with no non-epidemic simulations). Given the temperature structure used, there are likely fewer days with extreme temperatures leading to higher mortality in this Tmean region. The central point of the optimal survival temperature range (Tempc) consistently ranked high in all models, while other mortality and seasonality parameters and interactions varied between models. However, the variance unexplained by the two-way models was still more than 50%. Higher order interactions may allow for better characterization of the contribution of different parameters, but preliminary testing on this model with a related parameter set (Lord, unpublished data) indicated that a full model with all interactions was required to reach high levels of variance explained. This will require restricting the number of parameters varied and increasing the number of parameter sets, and it is unclear if it will substantially improve our understanding. The importance of interactions, particularly with temperature parameters, was also identified in a coinfection model [34], particularly variation in and interactions with the temperature parameters. This model also used a quadratic relationship between mortality and temperature and considered variations in the temperature function but did not explicitly vary the mortality parameters. By combining variation in the mortality parameters and the temperature function and using a sensitivity analysis that allowed for the exploration of interactions, we highlighted the potential role of temperature–mortality interactions in CHIKV dynamics.
Multiple studies have shown adult mortality in these mosquito species to be temperature dependent (e.g., [6,7,8]), and there was frequently an optimal temperature or range for survival. This optimal temperature region varied in both location and width, both within and between species.
Mosquito populations and temperature dependence in parameters varies between models of CHIKV and other arboviruses. Models have been used to estimate unknown parameters using a compartmental model, subsequently fitting the model to epidemiological data or mosquito abundance data, and then using the fitted models to estimate other parameters and investigate mosquito control (e.g., [35,36,37]). Mosquito mortality can be modeled as temperature dependent (e.g., [3,34,35]) or fixed (e.g., [36,37]). Chen et al. [35] used the same base data as in this model [8] but estimated daily mortality by first fitting a quadradic equation to the data, applying that equation to daily temperatures to estimate a daily temperature-dependent mortality rate and finally fitting a sinusoid function to the estimated daily rates. The model in Chen et al. [35] was used to fit mosquito trap data and infer additional parameters to estimate carrying capacity, so they did not perform the same type of sensitivity analysis as used here. Beyond natural mortality, control strategies can be simulated by increasing mortality at various time points or intervals. This increased mortality decreased the number of cases and shifted the peak time of epidemics in Rio de Janeiro [36]. Using a model to estimate the efficacy of control efforts, then using that efficacy to investigate different scenarios, the timing and number of control events affected the cumulative number of CHIKV cases in France [37].
These models show that the specifics of the mortality function can affect arbovirus dynamics. In the model reported here, this was primarily through interactions between these and other mortality or seasonality parameters, with similar interactions between parameters having effects on coinfection rates [34]. The piecewise mortality–temperature function used here could replicate a wider range of relationships but inherently exhibits a pattern of high mortality at extreme temperatures, a central region of optimal temperatures for survival and linear regions in between. This pattern is also exhibited in other functions such as complex polynomials or quadratic equations, as has been fit in other studies (e.g., [6,7,8,34,35]). However, while a complex polynomial may exhibit a better fit to specific data, the parameters are not easily related to biological concepts and so are difficult to extrapolate. However, consideration of other structures, particularly with non-linear regions, and additional variation between the two species, would be worthwhile. The previous analysis showed that variation between the two species affected transmission dynamics [3]; extended simulation sets and analyses including both types of variation would be valuable in understanding transmission dynamics.
The relationship between the likelihood of epidemics and the seasonal temperature curve was complex, likely non-linear and warrants further research. This should include other relationships between the mean and amplitude of the annual temperature cycle, as the function used here was developed from Florida data only. Understanding this interaction will be important for comparing outcomes from different locations. It is worth noting that this model is deterministic and compartmental and considers only daily temperatures. Agent-based models would be valuable to extend this analysis, to consider finer time scales and individual mosquitoes experiencing varied temperature regimes. However, the trade-off in complexity will limit some aspects of analysis.
Supplementary Materials
The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/v17111486/s1: Table S1. full parameter sets used to run the simulations; Table S2. correlation coefficients between parameters; Table S3. model output used in analysis; Table S4. results from two-way interaction models.
Funding
This research received no external funding.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
All data used in analysis are provided in the Supplementary Materials. Model code is available in the Supplementary Materials for Lord et al. [3].
Acknowledgments
Sara Farless and Daniel Post assisted with analysis, figures and references. The authors have reviewed and edited the output and take full responsibility for the content of this publication.
Conflicts of Interest
The author declares no conflicts of interest.
Abbreviations
The following abbreviations are used in this manuscript:
| CHIKV | Chikungunya virus |
| Adj. R2 | Adjusted R2 |
References
- Bellan, S.E. The importance of age dependent mortality and the extrinsic incubation period in models of mosquito-borne disease transmission and control. PLoS ONE 2010, 5, e10165. [Google Scholar] [CrossRef]
- Blower, S.M.; Dowlatabadi, H. Sensitivity and uncertainty analysis of complex models of disease transmission: An HIV model, as an example. Int. Stat. Rev. 1994, 62, 229–243. [Google Scholar] [CrossRef]
- Brady, O.J.; Johansson, M.A.; Guerra, C.A.; Bhatt, S.; Golding, N.; Pigott, D.M.; Delatte, H.; Grech, M.G.; Leisnham, P.T.; Maciel-de-Freitas, R.; et al. Modelling adult Aedes aegypti and Aedes albopictus survival at different temperatures in laboratory and field settings. Parasites Vectors 2013, 6, 351. [Google Scholar] [CrossRef]
- Braks, M.A.H.; Honorio, N.A.; Lourenco-De-Oliveira, R.; Juliano, S.A.; Lounibos, L.P. Convergent habitat segregation of Aedes aegypti and Aedes albopictus (Diptera: Culicidae) in southeastern Brazil and Florida. J. Med. Entomol. 2003, 40, 785–794. [Google Scholar] [CrossRef]
- CDC. Chikungunya in the United States. 2024. Available online: https://www.cdc.gov/chikungunya/data-maps/chikungunya-us.html (accessed on 31 July 2025).
- Chen, J.; Huo, X.; Wilke, A.B.B.; Beier, J.C.; Vasquez, C.; Petrie, W.; Cantrell, R.S.; Cosner, C.; Ruan, S. Linking mathematical models and trap data to infer the proliferation, abundance, and control of Aedes aegypti. Acta Trop. 2023, 239, 106837. [Google Scholar] [CrossRef]
- Dodero-Rojas, E.; Ferreira, L.G.; Leite, V.B.P.; Onuchic, J.N.; Contessoto, V.G. Modeling Chikungunya control strategies and Mayaro potential outbreak in the city of Rio de Janeiro. PLoS ONE 2020, 15, e0222900. [Google Scholar] [CrossRef]
- ECDPC. Chikungunya Virus Disease Worldwide Overview. 2024. Available online: https://www.ecdc.europa.eu/en/chikungunya-monthly (accessed on 31 July 2025).
- Garrett-Jones, C.; Shidrawi, G.R. Malaria vectorial capacity of a population of Anopheles gambiae. An exercise in epidemiological entomology. Bull. World Health Organ. 1969, 40, 531–545. [Google Scholar]
- Hopperstad, K.A.; Reiskind, M.H. Recent changes in the local distribution of Aedes aegypti (Diptera: Culicidae) in south Florida, USA. J. Med. Entomol. 2016, 53, 836–842. [Google Scholar] [CrossRef] [PubMed]
- Hopperstad, K.A.; Sallam, M.F.; Reiskind, M.H. Estimations of fine-scale species distributions of Aedes aegypti and Aedes albopictus (Diptera: Culicidae) in eastern Florida. J. Med. Entomol. 2021, 58, 699–707. [Google Scholar] [CrossRef] [PubMed]
- Jourdain, F.; de Valk, H.; Noël, H.; Paty, M.-C.; L’Ambert, G.; Franke, F.; Mouly, D.; Desenclos, J.-C.; Roche, B. Estimating chikungunya virus transmission parameters and vector control effectiveness highlights key factors to mitigate arboviral disease outbreaks. PLoS Neglected Trop. Dis. 2022, 16, e0010244. [Google Scholar] [CrossRef]
- Juliano, S.A.; Lounibos, L.P.; O’Meara, G.F. A field test for competitive effects of Aedes albopictus on A. aegypti in South Florida: Differences between sites of coexistence and exclusion? Oecologia 2004, 139, 583–593. [Google Scholar] [CrossRef] [PubMed]
- Kendrick, K.; Stanek, D.; Blackmore, C. Notes from the field: Transmission of Chikungunya virus in the continental United States--Florida, 2014. Morb. Mortal. Wkly. Rep. 2014, 63, 1137. [Google Scholar]
- Lord, C.C.; Day, J.F. Simulation studies of St. Louis encephalitis in South Florida. Vector Borne Zoonotic Dis. 2001, 1, 299–316. [Google Scholar] [CrossRef] [PubMed]
- Lord, C.C.; Day, J.F. Simulation studies of St. Louis encephalitis and West Nile viruses: The impact of bird mortality. Vector Borne Zoonotic Dis. 2001, 1, 317–330. [Google Scholar] [CrossRef] [PubMed]
- Lord, C.C.; Lounibos, L.P.; Pohedra, J.J.; Alto, B.W. Effects of mosquito biology on modeled Chikungunya virus invasion potential in Florida. Viruses 2020, 12, 830. [Google Scholar] [CrossRef]
- Lounibos, L.P.; Bargielowski, I.; Carrasquilla, M.C.; Nishimura, N. Coexistence of Aedes aegypti and Aedes albopictus (Diptera: Culicidae) in peninsular Florida two decades after competitive displacements. J. Med. Entomol. 2016, 53, 1385–1390. [Google Scholar] [CrossRef]
- Lounibos, L.P.; Kramer, L.D. Invasiveness of Aedes aegypti and Aedes albopictus and vectorial capacity for chikungunya virus. J. Infect. Dis. 2016, 214 (Suppl. S5), S453–S458. [Google Scholar] [CrossRef]
- Lwande, O.W.; Obanda, V.; Lindström, A.; Ahlm, C.; Evander, M.; Näslund, J.; Bucht, G. Globe-trotting Aedes aegypti and Aedes albopictus: Risk factors for arbovirus pandemics. Vector Borne Zoonotic Dis. 2020, 20, 71–81. [Google Scholar] [CrossRef] [PubMed]
- Morrison, T.E. Reemergence of chikungunya virus. J. Virol. 2014, 88, 11644–11647. [Google Scholar] [CrossRef]
- Ponlawat, A.; Harrington, L.C. Blood Feeding Patterns of Aedes aegypti and Aedes albopictus in Thailand. J. Med. Entomol. 2005, 42, 844–849. [Google Scholar] [CrossRef] [PubMed]
- Poterek, M.L.; Vogels, C.B.F.; Grubaugh, N.D.; Ebel, G.D.; Alex Perkins, T.; Cavany, S.M. Interactions between seasonal temperature variation and temporal synchrony drive increased arbovirus co-infection incidence. R. Soc. Open Sci. 2022, 9, 220829. [Google Scholar] [CrossRef] [PubMed]
- Reiskind, M.H.; Lounibos, L.P. Spatial and temporal patterns of abundance of Aedes aegypti L. (Stegomyia aegypti) and Aedes albopictus (Skuse) [Stegomyia albopictus (Skuse)] in southern Florida. Med. Vet. Entomol. 2012, 27, 421–429. [Google Scholar] [CrossRef] [PubMed]
- Richards, S.L.; Ponnusamy, L.; Unnasch, T.R.; Hassan, H.K.; Apperson, C.S. Host-feeding patterns of Aedes albopictus (Diptera: Culicidae) in relation to availability of human and domestic animals in suburban landscapes of central North Carolina. J. Med. Entomol. 2006, 43, 543–551. [Google Scholar] [CrossRef]
- Salami, D.; Capinha, C.; Sousa, C.A.; Martins, M.d.R.O.; Lord, C. Simulation models of dengue transmission in Funchal, Madeira Island: Influence of seasonality. PLoS NTD 2020, 14, e0008679. [Google Scholar] [CrossRef]
- Schmidt, C.A.; Comeau, G.; Monaghan, A.J.; Williamson, D.J.; Ernst, K.C. Effects of desiccation stress on adult female longevity in Aedes aegypti and Ae. albopictus (Diptera: Culicidae): Results of a systematic review and pooled survival analysis. Parasites Vectors 2018, 11, 267. [Google Scholar] [CrossRef]
- Scott, T.W.; Amerasinghe, P.H.; Morrison, A.C.; Lorenz, L.H.; Clark, G.G.; Strickman, D.; Kittayapong, P.; Edman, J.D. Longitudinal studies of Aedes aegypti (Diptera: Culicidae) in Thailand and Puerto Rico: Blood feeding frequency. J. Med. Entomol. 2000, 37, 89–101. [Google Scholar] [CrossRef]
- Scott, T.W.; Chow, E.; Strickman, D.; Kittayapong, P.; Wirtz, R.A.; Lorenz, L.H.; Edman, J.D. Blood-feeding patterns of Aedes aegypti (Diptera: Culicidae) collected in a rural Thai village. J. Med. Entomol. 1993, 30, 922–927. [Google Scholar] [CrossRef] [PubMed]
- Scott, T.W.; Morrison, A.C.; Lorenz, L.H.; Clark, G.G.; Strickman, D.; Kittayapong, P.; Zhou, H.; Edman, J.D. Longitudinal studies of Aedes aegypti (Diptera: Culicidae) in Thailand and Puerto Rico: Population dynamics. J. Med. Entomol. 2000, 37, 77–88. [Google Scholar] [CrossRef]
- Sharp, T.M.; Roth, N.M.; Torres, J.; Ryff, K.R.; Pérez Rodríguez, N.M.; Mercado, C.; Pilar Diaz Padró, M.D.; Ramos, M.; Phillips, R.; Lozier, M.; et al. Chikungunya cases identified through passive surveillance and household investigations—Puerto Rico, 5 May–12 August 2014. MMWR Morb. Mortal. Wkly. Rep. 2014, 63, 1121–1128. [Google Scholar]
- Smith, D.L.; Battle, K.E.; Hay, S.I.; Barker, C.M.; Scott, T.W.; McKenzie, F.E. Ross, Macdonald, and a theory for the dynamics and control of mosquito-transmitted pathogens. PLoS Pathog. 2012, 8, e1002588. [Google Scholar] [CrossRef]
- Staples, J.E.; Breiman, R.F.; Powers, A.M. Chikungunya fever: An epidemiological review of a re-emerging infectious disease. Clin. Infect. Dis. 2009, 49, 942–948. [Google Scholar] [CrossRef] [PubMed]
- Vazeille, M.; Moutailler, S.; Coudrier, D.; Rousseaux, C.; Khun, H.; Huerre, M.; Thiria, J.; Dehecq, J.-S.; Fontenille, D.; Schuffenecker, I.; et al. Two Chikungunya isolates from the outbreak of La Reunion (Indian Ocean) exhibit different patterns of infection in the mosquito, Aedes albopictus. PLoS ONE 2007, 2, e1168. [Google Scholar] [CrossRef] [PubMed]
- Vega-Rua, A.; Lourenco-de-Oliveira, R.; Mousson, L.; Vazeille, M.; Fuchs, S.; Yebakima, A.; Gustave, J.; Girod, R.; Dusfour, I.; Leparc-Goffart, I.; et al. Chikungunya virus transmission potential by local Aedes mosquitoes in the Americas and Europe. PLoS Neglected Trop. Dis. 2015, 9, e0003780. [Google Scholar] [CrossRef]
- Vega-Rúa, A.; Zouache, K.; Girod, R.; Failloux, A.B.; Lourenço-de-Oliveira, R. High level of vector competence of Aedes aegypti and Aedes albopictus from ten American countries as a crucial factor in the spread of Chikungunya virus. J. Virol. 2014, 88, 6294–6306. [Google Scholar] [CrossRef]
- Yang, H.M.; Macoris, M.L.G.; Galvani, K.C.; Andrighetti, M.T.M.; Wanderley, D.M.V. Assessing the effects of temperature on the population of Aedes aegypti, the vector of dengue. Epidemiol. Infect. 2009, 137, 1188–1202. [Google Scholar] [CrossRef] [PubMed]
Figure 1.
Simulation output. Two example runs (left panels: #1; right panels #219) shown illustrating two model behaviors, single outbreaks and bimodal outbreaks. Parameters for each run given in Table S1. Top panels: Hs, Hi, Hr: susceptible, infectious and recovered humans, respectively. Middle (Ae. albopictus) and bottom panels (Ae. aegypti): Sj: susceptible mosquitoes; Lj: latently infected mosquitoes; Yj, infectious mosquitoes. The red circle indicates the maximum infected host population, MaxHi, and the small arrow indicates the day the virus was introduced, tcrit.
Figure 1.
Simulation output. Two example runs (left panels: #1; right panels #219) shown illustrating two model behaviors, single outbreaks and bimodal outbreaks. Parameters for each run given in Table S1. Top panels: Hs, Hi, Hr: susceptible, infectious and recovered humans, respectively. Middle (Ae. albopictus) and bottom panels (Ae. aegypti): Sj: susceptible mosquitoes; Lj: latently infected mosquitoes; Yj, infectious mosquitoes. The red circle indicates the maximum infected host population, MaxHi, and the small arrow indicates the day the virus was introduced, tcrit.
Figure 2.
Effect of Tempc, Tmean and their interaction on epi, the likelihood of epidemics. Epi is binomial; yes = epidemic occurred (blue circles) and no = epidemic did not occur (brown triangles). Runs in a region around Tmean = 22 °C tended towards epidemic, with some influence of Tempc.
Figure 2.
Effect of Tempc, Tmean and their interaction on epi, the likelihood of epidemics. Epi is binomial; yes = epidemic occurred (blue circles) and no = epidemic did not occur (brown triangles). Runs in a region around Tmean = 22 °C tended towards epidemic, with some influence of Tempc.
Figure 3.
Effect of Tempc, on MaxHi, the peak number of hosts infected. Higher Tempc increases MaxHi, although the effect is small.
Figure 3.
Effect of Tempc, on MaxHi, the peak number of hosts infected. Higher Tempc increases MaxHi, although the effect is small.
Figure 4.
Annual mean temperature (Tmean) and day of introduction (tcrit) effect on MaxHi. Introductions later in the year with lower mean temperatures generated smaller outbreaks.
Figure 4.
Annual mean temperature (Tmean) and day of introduction (tcrit) effect on MaxHi. Introductions later in the year with lower mean temperatures generated smaller outbreaks.
Table 1.
State variables, parameters and parameter distributions used in the model and LHC analysis.
Table 1.
State variables, parameters and parameter distributions used in the model and LHC analysis.
| Descriptions | Symbol | Distribution | Range | Center |
|---|---|---|---|---|
| State variables | ||||
| Susceptible humans (no.) | Hs | |||
| Infectious humans (no.) | Hin | |||
| Recovered humans (no.) | Hr | |||
| Susceptible mosquitoes, species j (no.) | Sj | |||
| Latent mosquitoes, species j (no.) | Lj | |||
| Infectious mosquitoes, species j (no.) | Yj | |||
| Parameters | ||||
| Mean of temperature curve (°C) | Tmean | Uni | 18–27 | |
| Day virus is introduced (day) | tcrit | Uni | 5–360 | |
| Human recovery rate (day−1) | rH | Fixed | 0.125 | |
| Total (initial) number of humans (no.) | Htot | Fixed | 50,000 | |
| Both mosquito species | ||||
| Interval between pulses (both species) (days) | iv | Fixed | 20 | |
| Mean day peak one (from January 1) (day) | q1 | Fixed | 165 | |
| Spread peak one (days) | σ1 | Fixed | 7 | |
| Mean day peak two (from January 1) (day) | q2 | Fixed | 245 | |
| Spread peak two (days) | σ2 | Fixed | 15 | |
| Width of optimal survival range (either side of Tempc) (°C) | W | Uni a | 2–5 | 2 |
| Central point of optimal survival temperature range (°C) | Tempc | Tri b | 10–25 | 20 |
| Aedes albopictus | ||||
| Transmission mosquito to human host (day−1) | balb | Fixed | 0.5 | |
| Transmission human host to mosquito (day−1) | βalb | Fixed | 0.5 | |
| Days between blood feeding on humans (day) | αalb | Fixed | 5 | |
| Proportion of population in peak one (prop) | pδ1,alb | Fixed | 0.2 | |
| Proportion of population in pulses all year (prop) | pbase,alb | Fixed | 0.15 | |
| Minimum mortality (at Tempc ± W) (day−1) | μmin,alb | Tri | 0.04–0.1 | 0.06 |
| Temperature–mortality slope (mortality/Temp) | μsl,alb | Uni | 0.05–0.15 | |
| Virus development at 22.5 °C (day−1) | γ22,alb | Fixed | 0.25 | |
| Temperature–virus development slope (devel. × Temp−1) | γsl,alb | Fixed | 0.015 | |
| Total Albo recruitment through year (no.) | Rtot,alb | Tri | 100–20,000 | 5000 |
| Aedes aegypti | ||||
| Transmission mosquito to human host (day−1) | baeg | Fixed | 0.5 | |
| Transmission human host to mosquito (day−1) | βaeg | Fixed | 0.5 | |
| Days between blood feeding on humans (day) | αaeg | Fixed | 3 | |
| Proportion of population in peak one (prop) | p1,aeg | Fixed | 0.25 | |
| Proportion of population in pulses all year (prop) | pbase,aeg | Fixed | 0.13 | |
| Minimum mortality (at Tempc ± W) (day−1) | μmin,aeg | Tri | 0.04–0.1 | 0.08 |
| Slope of temperature–mortality line (mortality/Temp) | μsl,aeg | Uni | 0.05–0.15 | |
| Virus development at 22.5 °C (day−1) | γ22,aeg | Fixed | 0.25 | |
| Temperature–virus development slope (devel. × Temp−1) | γsl,aeg | Fixed | 0.015 | |
| Total Ae. aegypti recruitment through year (no.) | Rtot,aeg | Tri | 100–20,000 | 5000 |
a Uni = uniform distribution used; parameters are the low–high values (range). b Tri = triangular distribution used; parameters are the low–high values (range) and the most probable (center) value.
Table 2.
Main effects models (nine parameters) for epi and MaxHi. Top three parameters for each model indicated by bolded p-value.
Table 2.
Main effects models (nine parameters) for epi and MaxHi. Top three parameters for each model indicated by bolded p-value.
| Main Effects Model for MaxHi | ||
| Descriptions | Symbol | p Values |
| Central point of optimal survival temperature range | Tempc | 0.000006 |
| Total recruitment through year (Ae. albpopictus) | Rtot,alb | 0.000553 |
| Width of optimal survival range (either side of Tempc) | W | 0.005202 |
| Slope of temperature–mortality line (Ae. aegypti) | μsl,aeg | 0.036429 |
| Mean of temperature curve | Tmean | 0.250764 |
| Temperature–mortality slope | μsl,alb | 0.402013 |
| Day virus is introduced | tcrit | 0.543823 |
| Minimum mortality (at Tempc ± W) (Ae. albopictus) | μmin,alb | 0.690523 |
| Minimum mortality (at Tempc ± W) (Ae. aegypti) | μmin,aeg | 0.704036 |
| Main Effects Model for epi | ||
| Descriptions | Symbol | p Values |
| Central point of optimal survival temperature range | Tempc | 0.00692 |
| Slope of temperature–mortality line | μsl,aeg | 0.010205 |
| Day virus is introduced | tcrit | 0.032391 |
| Mean of temperature curve | Tmean | 0.679624 |
| Minimum mortality (at Tempc ± W) (Ae. aegypti) | μmin,aeg | 0.786395 |
| Minimum mortality (at Tempc ± W) (Ae. albopictus) | μmin,alb | 0.821233 |
| Temperature-mortality slope (Ae. albopictus) | μsl,alb | 0.852199 |
| Total recruitment through year (Ae. albopictus) | Rtot,alb | 0.868015 |
Table 3.
Top five parameters (highest p value) for main effects and interaction models for epi and MaxHi.
Table 3.
Top five parameters (highest p value) for main effects and interaction models for epi and MaxHi.
| Main Effects and Two-Way Interaction Model for MaxHi | ||||
| Descriptions | Predictor 1 | Predictor 2 | p Values | Main Effect/Two-Way Interactions |
| Mean of temperature curve Day virus is introduced | Tmean | tcrit | 0.008151 | Two-way interaction |
Temperature–mortality slope (Ae. albopictus) Center point of optimal temperature range | μsl,alb | Tempc | 0.052552 | Two-way interaction |
Mean of temperature curve | Tmean | Tempc | 0.053523 | Two-way interaction |
| Center point of optimal temperature range Center point of optimal temperature range | Tempc | 0.157706 | Main effect | |
Mean of temperature curve Temperature-mortality slope (Ae. albopictus) | Tmean | μsl,alb | 0.226489 | Two-way interaction |
| Main Effects and Two-Way Interaction Model for epi | ||||
| Descriptions | Predictor 1 | Predictor 2 | p Values | Main Effects/Two-Way Interactions |
| Mean of temperature curve Center temperature for mortality | Tmean | Tempc | 0.000001 | Two-way interaction |
Center temperature for mortality Width of optimal survival range (either side of Tempc) | Tempc | W | 0.000112 | Two-way interaction |
Center temperature for mortality | Tempc | 0.000141 | Main effect | |
| Day virus is introduced | tcrit | 0.011305 | Main effect | |
Mean of temperature curve Temperature–mortality slope (Ae. albopictus) | Tmean | μsl,alb | 0.01481 | Two-way interaction |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
MDPI and ACS Style
Lord, C.C. Effects of Temperature Dependence in Mosquito Mortality on Simulated Chikungunya Virus Transmission. Viruses 2025, 17, 1486. https://doi.org/10.3390/v17111486
AMA Style
Lord CC. Effects of Temperature Dependence in Mosquito Mortality on Simulated Chikungunya Virus Transmission. Viruses. 2025; 17(11):1486. https://doi.org/10.3390/v17111486
Chicago/Turabian StyleLord, Cynthia C. 2025. "Effects of Temperature Dependence in Mosquito Mortality on Simulated Chikungunya Virus Transmission" Viruses 17, no. 11: 1486. https://doi.org/10.3390/v17111486
APA StyleLord, C. C. (2025). Effects of Temperature Dependence in Mosquito Mortality on Simulated Chikungunya Virus Transmission. Viruses, 17(11), 1486. https://doi.org/10.3390/v17111486
Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.
