Reproduced, with permission, from:
Haile, D. G. 1989. Computer simulation of the effects of changes in weather patterns on vector-borne disease transmission. In The potential effects of global climate change in the United States, ed. J. B. Smith and D. A. Tirpak. Document no. 230-05-89-057, Appendix G. Washington, D.C.: U.S. Environmental Protection Agency.
D. G. Haile
Insects Affecting Man and Animals Research Laboratory
Agricultural Research Service
U. S. Department of Agriculture
Gainesville, FL 32604
Project No. DW12932662-01-1
Two weather-based models for simulation of the population dynamics of disease vectors were used to assess the possible effects of climate change due to increased levels of CO2 in the atmosphere on vector-borne disease transmission in the United States. The first model (ADTSIM) simulates population dynamics of the American dog tick (ADT), Dermacentor variabilis, which is the primary vector of Rocky Mountain Spotted Fever (RMSF) in the eastern U. S. This model included the effects of temperature and atmospheric moisture on the life processes of the tick. The density of adult ticks was used as an indicator of RMSF transmission potential because this is the stage normally involved in transmission of human cases. The second model (MALSIM) simulates the population dynamics of Anopheline mosquitoes and the transmission of malaria between mosquitoes and humans. This model simulates direct incidence of malaria, assuming that the disease is reintroduced into a human population that is continuously exposed to mosquito bites. The effects of temperature, atmospheric moisture, and rainfall are included in MALSIM. The malaria vector considered in these simulations is Anopheles quadrimaculatus, which presently exists in many areas of the eastern half of the U. S. and was the primary vector of malaria in the south and east when the disease was endemic.
The ADTSIM results indicate that with the proposed climatic change scenarios at certain southern locations (Jacksonville, FL, and San Antonio, TX) ADT populations will disappear owing to adverse effects of high temperatures and low humidity, while at certain northern locations (Missoula, MT, North Bay, Ont., and Halifax, N.S.) populations will increase because of warming with adequate moisture. For most other U. S. locations, tick densities either declined moderately or remained the same with various weather scenarios. These results suggest that overall in the U. S. the problem of RMSF will decrease in certain areas, while others will remain unchanged.
The results of simulations of malaria transmission (assuming the disease is reintroduced into an unprotected population) show that there is little change in the transmission potential with modified weather scenarios. Areas in Florida with high transmission potential normally will remain high; very little increase was predicted for other areas.
Only the direct effects of weather on the disease vector were considered in these simulations. No consideration of the effects of weather on host densities or habitat were included. Consideration of these effects, along with other model refinements, would be necessary to improve confidence in the overall results. Also, much additional research on the biology, ecology, and modeling of these and other vector/disease complexes will be required for a more complete analysis of the impact of weather change on vector-borne diseases.
This study addresses the impact of climate change on vector-borne disease transmission in the United States as part of an overall study by the U. S. Environmental Protection Agency (EPA) the effects of climatic change resulting from the "greenhouse effect". Although at present there are no vector-borne diseases that are continuously transmitted to the human population in the United States, numerous arthropod vectors are present in the environment and the potential for major outbreaks exists in various regions as a result of increases in vector populations or prevalence of a disease.
Certain diseases are present in animal hosts, and others may be introduced from endemic areas by human immigrants or travelers. The arthropods most commonly involved in disease transmission in the United States involve various tick and mosquito species. The importance of weather variables on vector population dynamics and disease transmission has long been recognized (Harwood and James, 1979; Warren and Mahoud, 1984; Russell et al., 1963). Increases in temperature, up to an optimum level, normally increases the development and survival rate of vector and parasite life stages. Adequate atmospheric or soil moisture levels are important for survival of various vector stages and, in the case of mosquitoes, adequate levels of aquatic habitat are necessary for immature development.
Models have been used as a tool for understanding various aspects of vector population dynamics and disease transmission. In particular, the transmission of malaria has been studied extensively with mathematical models (Ross, 1911; MacDonald, 1957; Deitz et al., 1984; Molineaux and Grammiccia, 1980). Some efforts have been made to study mosquito and tick population dynamics using a computerized life history approach (Haile and Weidhaas, 1977; Fine et al., 1979; Sutherst et al., 1978; Sonenshine, 1975), but none of these efforts attempted to comprehensively model the effects of weather variables required for a quantitative assessment of climatic change. More recently, however, a comprehensive computer simulation model for population dynamics of the lone star tick, Amblyomma americanum, was developed, which included the direct effects of weather variables (Haile and Mount, 1987). This study was extended to develop a similar model for population dynamics of the American dog tick, Dermacentor variabilis, which is the principal vector of Rocky Mountain Spotted Fever (RMSF) in the U. S. (Mount and Haile, in press). In addition, the model for Anopheles albimanus by Haile and Weidhaas (1977) has been expanded to include the direct effects of weather variables and to allow simulation of malaria transmission to and from a human population with a variety of vector species (unpublished). These weather-sensitive models were used as the basis for the study on the impact of climatic change reported here. This report will describe the methodology used to simulate the effects of climatic change and an analysis of the implications of the results.
THE AMERICAN DOG TICK MODEL
The development and validation of the American dog tick model (ADTSIM) is reported in Mount and Haile (in press). Briefly, this model uses a dynamic life history approach with weekly age classes and time steps. Environmental variables used in the model include temperature, saturation deficit, daylength, host density, and habitat type. Relationships between environmental and biological variables include the following: (1) temperature-dependent development rates for eggs and engorged larvae, nymphs, and females; (2) the influence of temperature on fecundity; (3) the influence of habitat type, temperature, and saturation deficit on survival rates of free-living ticks; (4) the effect of host density, temperature, and daylength on host-finding rates; and (5) density-dependent survival of parasitic ticks during engorgement.
ADTSIM was developed to reflect the effects of average weekly weather data. These data may be actual values for a particular year or historical average data (values averaged over a period of years). As such the model was not designed to reflect the influence of extreme conditions on a diurnal or daily basis.
The effect of climate change in ADTSIM will indicate the direct effects of changes in ambient temperature and moisture on the tick life cycle with all other variables remaining constant. Any effect of climate change on habitat type and host densities are not simulated. An average of the number of adult ADT on hosts each week was used as an index to summarize each year of simulation. This index can also be used as an indicator of the transmission potential of RMSF because the adult tick is most often involved in transmission of the disease to humans.
Weather scenarios based on the results of three climate change models (GISS, GFDL, and OSU) that simulated the effects of doubling the level of CO2 in the atmosphere, were used in the vector population models. Only simulations representing a step change in weather were used in this study. Actual data for the base period 1951-1980 were used for only one location (Richmond, VA) with ADTSIM. Historical average weather data for Richmond and other locations were used as the basis for evaluating the effect of the weather change models at each location. In this case the weekly weather data (normal or modified) were used in the population model each year until equilibrium was reached (usually ca. 15 years). A comparison of the equilibrium population with the population resulting from use of modified weather gave a direct measure of the impact of climate change.
THE MALARIA MODEL
A comprehensive model (MALSIM) to simulate the effects of environmental variables on population dynamics of Anopheline mosquitoes and on transmission of Falciparum malaria to a human population is currently being developed. Although this model is not fully validated, it contains the necessary relationships to provide a preliminary evaluation of the impact of climate change scenarios on this important disease.
Anopheles quadrimaculatus is the only malaria vector simulated in these studies. This species is still present in many areas of the eastern half of the United States and was the primary vector of malaria in the Southeast when the disease was endemic.
The following environmental relationships are included in MALSIM at present: (1) the effect of temperature on developmental rates of each vector stage and the parasite in the mosquito; (2) the effect of temperature on survival of immature stages; (3) the effect of temperature and saturation deficit on adult survival; (4) the relationship between rainfall, available area of aquatic habitat, immature density, and immature survival; and (5) temperature-induced hibernation of mated, nonblooded females.
Historical average weather data for a number of locations were used as the basis for simulations to compare the malaria transmission potential for normal vs. modified weather resulting from three climate change models (GISS, GFDL, and OSU). Although malaria is unlikely to become endemic in the United States, these simulations assume an introduction of 100 cases in an unprotected population of 10,000 along with initiation of mosquito population with 100,000 female adults. The level of transmission (incidence) in the 10,000 unprotected population after 3 years of simulation is used as an indication of transmission potential for comparative purposes.
ADTSIM RESULTS
ADTSIM - Richmond, Virginia 1950-80
The simulated adult tick densities for the period 1950-1980 at Richmond with normal yearly weather data shows considerable variation between years as a result of weather alone (Figure 1). Cases of RMSF in Virginia and the United States (CDC data) are plotted in Figure 1 to show the correlation between simulated tick density and actual disease incidence. Figure 2 shows the comparison between the simulations with normal weather and the three models with modified weather. All modified weather scenarios produce a decrease in tick density for all years in the 30-year period. The OSU and GISS models show very similar results while the GFDL model produced a greater reduction. Comparisons of the average density with yearly data to the equilibrium density with historical average weather data showed that the results are very similar. Therefore, the equilibrium density method was used for evaluation of the impact of the modified weather scenarios at other locations.
ADTSIM Results at Various Locations
The equilibrium density of ADT adults on hosts at various U.S. locations and two Canadian locations with normal weather compared to the modified weather scenarios is shown in Figure 3. The results indicate a very clear shift in tick densities from south to north. The tick population was eliminated at the most southerly locations (Jacksonville and San Antonio) for all three modified weather scenarios. Simulations at Albuquerque showed no population with the normal dry weather and no change with any of the scenarios. Robust increases in tick density were indicated at the most northerly locations (Missoula, North Bay, and Halifax) for all scenarios. Missoula and North Bay had no tick population with normal weather and Halifax had only a low-level population. Simulations at U.S. locations where the normal tick density is relatively high produced a variety of results for the different scenario models. At some locations the three models produced similar results showing essentially no change in density at Los Angeles, Boston, and Medford or a moderate decrease at Richmond, New York, and Columbus. The results at Tulsa and Nashville indicated a decline in populations, with the OSU and GISS models predicting moderate declines and the GFDL model predicting population elimination. Simulation at Minneapolis gave the most mixed results between the different scenarios, with a population increase for GISS, no change for OSU, and population elimination for GFDL. In general, for locations where simulation indicated a moderate decline in population, the GFDL model showed the greatest decline.
RESULTS WITH MALSIM
Simulated levels of malaria transmission potential at various locations in the United States with normal weather and the three modified weather scenarios are presented in Figure 4. These results show that for the locations with the highest normal malaria transmission potential (Miami, Key West, and Orlando) there is very little change for any of the modified weather scenarios. For Jacksonville, which has a medium transmission level with normal weather, the results of the scenario models were mixed, with the GFDL model showing an increase, no change for GISS, and a decrease for OSU. As a practical matter, the changes at Jacksonville appear to be of little importance. At other locations with normally low transmission potential, only the GISS model appeared to result in significant increases in transmission potential. The GISS model produced increases at San Antonio, Atlanta, Nashville, and Richmond, with the largest increase at Atlanta. Simulations at Tulsa, Dallas, Baltimore, Indianapolis, and Boston indicated extremely low normal transmission with no significant change with any modified weather patterns. Overall, these simulations suggest that the proposed climatic changes will have little if any impact on transmission potential of malaria in the United States.
INTERPRETATION AND IMPLICATIONS OF RESULTS
These results suggest that potential problems associated with transmission of RMSF and malaria in the United States will be little worse with projected climatic change than they are today. However, these are only two of many vector-borne diseases that are a potential problem in the United States. Other modeling efforts would be required to address other tick or mosquito-borne diseases.
The results in this report are based on computer models, which although complex and useful from a research standpoint, have serious limitations for projecting the effects of climatic change. The limitations involve development of the models with average weather data and the lack of consideration of the climatic effects on host densities and habitat. Additional research on biology, ecology, and modeling of these and other vector/disease systems will be required for a more complete analysis of the potential effects of climate change.
REFERENCES
Dietz, K., L. Molineaux, and A. Thomas. A malaria model tested in the African savannah. Bull. WHO, (50):347 57, 1984.
Fine, P.E.M., M.M. Milbey, and W.C. Reeves. A general simulation model for genetic control of mosquito species that fluctuate markedly in population size. J. Med. Entomol., (16):189-199, 1979.
Haile, D.G. and D.W. Weidhaas. Computer simulation of mosquito populations (Anopheles albimanus) for comparing the effectiveness of control techniques. J. Med. Entomol., (13):553-567, 1977.
Haile, D.G. and G A. Mount. Computer simulation of population dynamics of the lone star tick, Amblyomma americanum (Acari: Ixodidae). J. Med. Entomol. (24):356-369, 1987.
Harwood, R.F. and M.T. James. Entomology in Human and Animal Health. MacMillan Publishing Co. Inc., New York. 1979.
MacDonald, G. The Epidemiology and Control of Malaria. Oxford University Press, London. 1957. 201 pp.
Molineaux, L. and G. Grammiccia. The Garki Project: Research on the epidemiology and control of malaria in the Sudan savanna of West Africa World Health Organization, Geneva, Switzerland. 1980.
Mount, G A. and D.G. Haile. Computer simulation of population dynamics of the American dog tick, Dermacentor variabilis (Acari: Ixodidae). J. Med. Entomol. (in press)
Ross, R. The Prevention of Malaria, (2nd edition). London: Murray. 1911.
Russell, P.F., L.S. West, R.D. Manwell, and G. MacDonald. Practical Malariology, (2nd edition). Oxford Univeristy Press, London 1963.
Sonenshine, D.E. Influence of host-parasite interactions on the population dynamics of ticks. Misc. Pub. Entomol. Soc. Am. (9):243-249, 1975.
Sutherst, R.W., R.H. Wharton, and K.B.W. Utech. Guide to studies on tick ecology, CSIRO Aust. Div. Entomol. Tech. Pap. 14. 1978.
Warren, K.S. and A A.F. Mahmoud. 1984. Tropical and Geographical Medicine. McGraw-Hill Book Company, New York, 1984.
