Decision Support System For Mosquito and Arbovirus Control in California

EarthzineArticles, Biodiversity, Health, Original, Sections

The Culex tarsalis mosquito is an important vector of WNV to humans in the western US.

Fig. 1. The Culex tarsalis mosquito is an important vector of WNV to humans in the western US.

CM Barker1 , VL Kramer2 , WK Reisen1


California has over a century of experience with mosquito control and has studied arboviruses since their discovery as veterinary and public health problems during the 1930s. For the past 40 years, a statewide surveillance and intervention program coordinated by the California Department of Public Health has been implemented by local vector control agencies and supported by the University of California. The arrival of West Nile virus during the past decade has required enhanced control driven by expanded surveillance. The current paper focuses on two recent advancements: 1) the California Mosquito-borne Virus Surveillance and Response Plan, which estimates transmission risk to humans and guides intervention responses; and 2) the California Surveillance Gateway, which provides centralized storage of California’s surveillance data and facilitates rapid reporting and visualization of surveillance results. We also consider future improvements to the program, including the development of forecasting models for mosquito abundance and arbovirus transmission.


California has had a long history of mosquito control, beginning in 1904 with the successful control of pest mosquitoes in San Francisco Bay Area salt marshes through water drainage and oiling (Reeves 1985), followed by successful campaigns against malaria that began in 1909 (Gray and Fontaine 1957). The passage of the Mosquito Abatement Act in 1915 established publicly funded special districts to control mosquitoes (Gray 1912; Gray and Fontaine 1957; Reeves 1985). Early in the 1900s, growing agricultural interests in the San Joaquin Valley of California were threatened by major outbreaks of encephalomyelitis in equines (Reeves et al. 1990), with clinical symptoms identical to those seen during outbreaks in the central and western United States during the previous 60 years. This led to the isolation and identification of western equine encephalomyelitis virus (WEEV) (Meyer et al. 1931) and soon after, horses were protected by annual vaccination (Reeves and Hammon 1962). Because the majority of cases of this newly described virus occurred during the summer when biting insects were most abundant, researchers suspected that the causative agent might be carried by mosquitoes. Other arboviruses were discovered around the same time, including St. Louis encephalitis virus [SLEV; (Muckenfuss et al. 1934)], generating considerable interest in the epidemiology of these diseases.

Both viruses also were found to cause human illness in the Central Valley of California (Howitt 1938; Howitt 1939). A widespread epizootic attributed to WEEV in the western and central United States during 1941 and encephalitis epidemics caused by WEEV and SLEV in the Yakima Valley each summer from 1939-1942 led the Washington State Health Department to request an investigation by researchers at the University of California.

The studies that followed definitively identified WEEV and SLEV as causative agents of human encephalitis (Reeves et al. 1990) and used serological evidence to demonstrate that both wild and domestic birds and mammals were commonly exposed to WEEV and SLEV, with antibody prevalence among wild vertebrates higher in birds than in mammals (Hammon et al. 1941a). These Yakima Valley studies also yielded the first isolates of WEEV and SLEV from field-collected mosquitoes (Hammon et al. 1941b), which was a big step toward incriminating them as vectors of these viruses. Culex tarsalis (Fig. 1), a species that feeds frequently on birds, had a relatively high prevalence of infection, and serologic evidence that birds had higher antibody prevalence than mammals pointed toward a bird-Culex-bird infection cycle.

Fig. 2. Temperature-related risk depicted for mosquito agencies within California using temperature data from the NASA Terrestrial Observation and Prediction System.

Fig. 2. Temperature-related risk depicted for mosquito agencies within California using temperature data from the NASA Terrestrial Observation and Prediction System.

Once WEEV and SLEV had been identified as causative agents of human and equine illness and their transmission mechanisms described, surveillance was initiated to monitor the enzootic of activity of these viruses.

An extensive epidemic of neurological disease centered in the southern San Joaquin Valley of California occurred during the summer of 1952 (Longshore et al. 1956), redirecting mosquito control efforts and causing increased recognition of the importance of ecological events preceding the outbreak (Reeves and Hammon 1962). Fragmented local surveillance and control activities and research programs supported by the University of California soon were coordinated to form a statewide program under the direction of the California Department of Public Health (Longshore 1960), with the goal of providing an early warning system for future outbreaks to enable mosquito control to limit amplification and prevent human disease. Guidelines for the surveillance and control of mosquito borne encephalitis viruses were written at the state level (Reisen 1995; Walsh 1987).

The importance of coordinated surveillance of enzootic transmission to direct intervention was recognized again at the federal level in response to the extensive West Nile virus (WNV) epidemic (Moore et al. 2002). The ongoing WNV epidemic has been the largest mosquito borne encephalitis outbreak documented in North America and the largest WNV outbreak ever recorded globally (Kramer et al. 2008). The invasion of California by WNV in 2003 (Reisen et al. 2004) was followed by extensive wildlife, equine and human disease, expanded surveillance to direct and justify emergency control, and new public scrutiny of mosquito control programs. Expanded surveillance programs required enhanced data management, spatial and temporal visualization, and computational tools to assess risk and recommend appropriate intervention. Preliminary studies expanded Reeves’ observations (Reeves and Hammon 1962) on the importance of winter precipitation and spring temperatures on summer mosquito abundance and arbovirus risk (Barker et al. 2008; Barker 2008; Barker et al. 2009; Reisen et al. 2008; Wegbreit and Reisen 2000), laying the foundation for predictive models using remotely sensed variables.

Surveillance Components

The North American encephalitis viruses, including WNV, are maintained in nature in a basic transmission cycle involving culicine mosquitoes and birds in the order Passeriformes. During favorable environmental conditions, transmission of these viruses may amplify to levels where spillover to other vertebrates such as equines and humans becomes likely. Infections in these tangential hosts may cause serious disease, but does not result in further virus transmission. WNV infection also has had a devastating effect on some avian species, especially those in the family Corvidae (Wheeler et al. 2009). Surveillance programs within California monitor enzootic transmission within the basic maintenance cycle to provide an early warning of viral amplification in sufficient time for intervention to prevent human disease (Reisen and Barker 2008). The factors currently measured to assess the risk of viral transmission to humans in California are summarized below, and the level of each is translated into a risk score from 1-5. These scores are averaged to obtain a measure of overall risk to guide interventions. Although equine cases are monitored using a passive case detection system, most equines in California now have been intentionally or naturally immunized against both WEEV and WNV and therefore generally are not useful as a surveillance indicator of enzootic amplification.

Fig. 3.   Summary of thresholds for individual surveillance indicators used in calculating the California Mosquito-Borne Virus Surveillance and Response Plan risk levels.  Ranges for overall risk levels are shown in the small box.

Fig. 3. Risk levels for WNV related to temperature. Note that risk levels increase as the duration of the median extrinsic incubation period (EIP) decreases and the rate of virus development (1/EIP) increases (Reisen et al. 2006).

1. Temperature. Temperature surfaces for California are acquired at 1 km2 resolution from NASA’s Terrestrial Observation and Prediction System (TOPS) (Fig. 2) and scored into quintiles based on the extrinsic incubation period of WNV within the Culex mosquito vector estimated from laboratory infection experiments (Figs. 2, 3).

In general, areas with the warmest temperatures and thus shortest extrinsic incubation periods have the highest risk of virus transmission.

2. Mosquito abundance. Culex vector species are collected by light (Mulhern 1953), dry ice-baited (Newhouse et al. 1966) and gravid female (Cummings 1992) traps and enumerated by species, sex and physiological condition. Female mosquitoes per trap type per night per collection period are expressed as a percentage of previous 5-year averages to assess population abundance and are interpreted as a percentage anomaly. Risk scores are assigned as below–above average based on the amplitude of the anomaly.

3. Mosquito virus infection. Mosquitoes collected alive by dry ice-baited or gravid traps are allocated into lots (i.e., pools) of ‰ä_50 females and then tested for virus infection using a real time RT-PCR triplex assay that tests for WEEV, SLEV and WNV. Infection prevalence is expressed as infections per 1,000 females by species, location, and time period using maximum likelihood estimation (Biggerstaff 2003) that accounts for variations in pool size. Risk levels increase with increasing infection prevalence.

4. Dead birds (WNV only). For each geographic area, the occurrence of laboratory-confirmed WNV-positive dead birds provides an indication of risk, and scores are assigned based on the geographic proximity and number of positives. Dead birds are reported by the public and shipped to the California Animal Health and Food Safety (CAHFS) laboratory for necropsy. Oral swabs and/or kidney snips are tested for WNV using a singleplex RT-PCR.

5. Sentinel chickens. Typically, flocks of 10 hens are deployed at ca. 230 locations throughout the state and then bled biweekly to detect infection with WEEV, SLEV or WNV by screening for antibodies using an enzyme immunoassay (EIA) (Reisen et al. 1994). Positives are confirmed and the infecting virus is identified by IFA, Western blot or plaque reduction neutralization test (Patiris et al. 2008). Dead or antibody-positive chickens are replaced with sero-negative birds. Risk is assigned based on the extent and intensity of virus transmission indicated by the numbers of positive flocks and birds per flock, respectively.

6. Human cases. The passive surveillance system monitoring human disease cases often has significant delays in reporting and limitations in the disclosure of infection location due to HIPAA restrictions. Problems and delays with diagnosis and reporting are compounded in California where an ethnically diverse human population of >38 million utilizes a mosaic of medical providers and diagnostic laboratories, creating geographic, jurisdictional, and linguistic challenges. In addition, recent large-scale influenza outbreaks have caused health providers to become reluctant to test for WNV-related febrile illness, confounding the counts of human cases as an indication of tangential transmission (Kwan et al. 2010b). For these reasons, human cases are not used generally for assessment of viral transmission risk but rather as an outcome measure to evaluate risk calculations (Barker et al. 2008).

Response Plan and the Surveillance Gateway

Small box with Risk levels.

The current California Mosquito-borne Virus Surveillance and Response Plan (CMBVSRP) (Kramer 2009) utilizes the monitoring data described above, which are scored from 1-5 and then averaged to calculate enzootic risk as normal season (1.0 – 2.5), emergency planning (>2.5 ‰ÛÒ 4.0) or epidemic (>4.1 ‰ÛÒ 5.0) level for each of the three viruses. Previously, these measures of risk agreed well with the occurrence of arboviral disease in humans during seasons with no, enzootic, and epizootic transmission of WEEV and SLEV (Barker et al. 2003). The threshold values used for each of the parameters used for estimating risk for WNV are summarized in Fig. 4 and relate directly to guidelines for intervention that escalate proportionally to increasing risk (Kramer 2009).

Fig. 4.  Summary of thresholds for individual surveillance indicators used in calculating the California Mosquito-Borne Virus Surveillance and Response Plan risk levels.  Ranges for overall risk levels are shown in the small box.

Fig. 4. Summary of thresholds for individual surveillance indicators used in calculating the California Mosquito-Borne Virus Surveillance and Response Plan risk levels. Ranges for overall risk levels are shown in the small box.

For surveillance data to be useful in intervention decision support, it must be acquired rapidly, analyzed quickly and presented in easy to understand visualizations. Data management in California has been greatly enhanced and expedited by the almost universal use of the California Vector borne Disease Surveillance Gateway (CalSurv Gateway; by public health and vector control agencies. This internet-based data acquisition, storage, retrieval, analysis, and visualization tool provides user-level security and includes a variety of mapping and reporting features useful in decision support. All sampling sites are registered at first use and include geographic coordinates that can be specified by the user or located visually using an interactive Google Maps tool. All data are stored in an underlying PostgreSQL database with PostGIS for advanced data retrieval and aggregation in both space and time.

Redundant data entry and errors are avoided, because enzootic surveillance data are entered once at the point of collection by participating agencies and test results are added to the same data source by supporting laboratories testing mosquitoes, dead birds and sentinel chickens. Errors in data entry are minimized by comprehensive and consistent data definition and the use of drop-down menus in the CalSurv Gateway’s ‘front-end’ user interface. Rapid laboratory diagnostics such as a multiplex qRT-PCR for viral RNA and EIA for antibodies allow specimens to be collected during Mon-Tues and the test results completed and available by Thurs-Fri of the same week. Once laboratory results are entered, reports are sent by auto generated email to the submitting agency, state-wide maps refreshed, and data sets updated. At weekly intervals, reports are auto generated to California Department of Public Health and data collated in .xml format for entry into the CDC nationwide ArboNET reporting system. In addition, half-monthly summaries of arboviral transmission risk are sent automatically in an e-mailed PDF to each participating agency, including an assessment of overall risk and the levels of each risk factor (Fig. 5). Estimates of risk are delimited by local agency surveillance efforts.

Future Plans

California’s current system for assessing arboviral transmission risk is based on the intensive data collection efforts of local vector control agencies and supporting laboratories. Early antecedent forecasting of pending risk of arbovirus transmission levels should allow vector control agencies to extend control efforts, limit adult vector abundance, and thereby preclude spillover transmission to humans (Reisen and Barker 2008). Forecasted risk is ‘truthed’ in real-time by surveillance that tracks risk measured as described above. Our current work aims to build suitable models using a blend of statistical and mechanistic approaches that accurately predict virus activity seasons in advance. Reasonable accuracy is paramount for the acceptance and response to these predictions by local agencies, because health education and intervention campaigns lose momentum if predictions of imminent threats are not realized. Preliminary analyses have shown that winter climate variation was well correlated with measures of mosquito abundance during the same and subsequent seasons (Barker 2008; Reisen et al. 2008). The entomological aspects of arboviral transmission may be expressed by vectorial capacity (Garrett-Jones 1964; Reisen 1989), which indicates that changes in vector abundance should be linearly related to virus transmission. Recently, increases in Culex tarsalis abundance weeks in advance were found to be positively associated with seroconversions to WEEV in sentinel chickens during the 1990s (Barker et al. 2010) and to WNV in sentinel chickens in Los Angeles from 2004-2008 (Kwan et al. 2010a). Interestingly, sentinel seroconversions in Los Angeles were detected concurrent with the onset of human cases, suggesting that this measure may monitor spillover transmission and thereby serve as a surrogate for passively detected human infections. In addition, vernal seroprevalence levels in populations of amplifying hosts seem to exert a negative impact on virus amplification (Kwan et al. 2010b) and result in a reduced risk of seroconversion in sentinel chickens (Kwan et al. 2010a).

Fig. 5. Example showing risk levels (range=1-5) per half-month for Kern Mosquito and Vector Control District during 2009 as normal season (yellow), emergency planning (orange) and epidemic (red) conditions. Overall risk levels for each species are indicated within the black box, and lower graphs present risk levels for each of the factors contributing to the overall risk.

Fig. 5. Example showing risk levels (range=1-5) per half-month for Kern Mosquito and Vector Control District during 2009 as normal season (yellow), emergency planning (orange) and epidemic (red) conditions. Overall risk levels for each species are indicated within the black box, and lower graphs present risk levels for each of the factors contributing to the overall risk.

Although mathematical (Bowman et al. 2005; Lewis et al. 2006; Wonham et al. 2004) and statistical (Bouden et al. 2008; Trawinski and Mackay 2008) models have been published for WNV, they have not been used extensively or operationally. We currently plan to extend these modeling efforts and incorporate climate variation, landscape measures and immunity in avian populations into predictions of WNV activity, perhaps depicted as risk surfaces for the State of California.


We especially thank the Kern Mosquito and Vector Control District and the many other participants in the California Encephalitis Virus Surveillance Program from the Mosquito and Vector Control Association of California, the California Department of Public Health and the Center for Vector borne Diseases at University of California, Davis, especially Programmer BK Park who wrote and maintains the California Surveillance Gateway web interface. This research was funded, in part, by cooperative agreement 1U01EH0001418 from the National Center for Environmental Health, CDC; Research Grant AI55607-A02 from the National Institute of Allergy and Infectious Diseases, NIH; the Climate Variability and Human Health program, Office of Global Programs, NOAA; Research Opportunities in Space and Earth Science Decision Support through Earth-Sun Science Research Results, NASA; and Research and Policy in Infectious Disease Dynamics (RAPIDD) Program, Fogarty Center, NIH and Department of Homeland Security.

1 Center for Vectorborne Diseases, Department of Pathology, Microbiology, and Immunology, School of Veterinary Medicine, University of California, Davis, CA 95616,

2Vector-Borne Disease Section, California Department of Public Health, Sacramento, CA

Reference List

Barker, C. M. 2008. Spatial and temporal patterns in mosquito abundance and virus transmission in California. PhD.University of California, Davis.

Barker CM, Johnson WO, Eldridge BF, Park BK, Melton F, Reisen WK. 2010. Temporal connections between Culex tarsalis abundance and transmission of western equine encephalomyelitis virus in California. Am J Trop Med Hyg 82:1185-1193.

Barker CM, Park B, Melton F, Eldridge BF, Kramer VL, Reisen WK. 2008. 2007 year-in-review: Integration of NASA’s meteorological data into the California response plan. Proc. Mosq. Vector Control Assoc. Calif. 76:7-12.

Barker CM, Reisen WK, Eldridge BF, Park B, Johnson WO. 2009. Culex tarsalis abundance as a predictor of western equine encephalomyelitis virus transmission. Proc. Mosq. Vector Control Assoc. Calif. 77:65-68.

Barker CM, Reisen WK, Kramer VL. 2003. California State Mosquito-borne Virus Surveillance and Response Plan: retrospective evaluation using conditional simulations. Am. J. Trop. Med. Hyg. 68:508-518.

Biggerstaff, BJ. 2003. Pooled infection rate. http://www. cdc. gov/ncidod/dvbid/westnile/software. htm : 1-5.

Bouden M, Moulin B, Gosselin P. 2008. The geosimulation of West Nile virus propagation: a multi-agent and climate sensitive tool for risk management in public health. Int. J. Health Geogr. 7:35

Bowman C, Gumel AB, van den DP, Wu J, Zhu H. 2005. A mathematical model for assessing control strategies against West Nile virus. Bull. Math. Biol. 67:1107-1133.

Cummings RF 1992. Design and use of a modified Reiter gravid mosquito trap for mosquito-borne encephalitis surveillance in Los Angeles County, California. Proc. Mosq. Vector Control Assoc. Calif. 60:170-176.

Garrett-Jones C 1964. Prognosis for interruption of malaria transmission through assessment of the mosquito’s vectorial capacity. Nature 204:1173-1175.

Gray HF 1912. Malaria control in California. Am. J. Hyg. 2:452-455.

Gray HF, Fontaine RE. 1957. A history of malaria in California. Proc. Calif. Mosq. Control Assoc. 25:18-39.

Hammon WM, Gray JA, Evans FC, Izumi EM. 1941a. Western equine and St. Louis encephalitis antibodies in the sera of mammals and birds from an endemic area. Science 94:305-307.

Hammon WM, Reeves WC, Brookman B, Izumi EM. 1941b. Isolation of viruses of Western equine and St. Louis encephalitis from Culex tarsalis mosquitoes. Science 94:328-330.

Howitt BF 1938. Recovery of the virus of equine encephalomyelitis from the brain of a child. Science 88:455

Howitt BF 1939. Viruses of equine and St. Louis encephalitis in relationship to human infections in California 1937 – 1938. Am. J. Publ. Hlth. 29:1083-1097.

Kramer LD, Styer LM, Ebel GD. 2008. A Global Perspective on the Epidemiology of West Nile Virus. Annu. Rev. Entomol. 53:61-81.

Kramer,VL. 2009. California State mosquito-borne virus surveillance and response plan. http://westnile. ca. gov/resources. php

Kwan JL, Kluh S, Madon MB, Nguyen DV, Barker CM, Reisen WK. 2010a. Utility of sentinel chicken seroconversions for West Nile virus surveillance in Los Angeles, California, 2004-2008. Am. J. Trop. Med. Hyg. [submitted]

Kwan JL, Kluh S, Madon MB, Reisen WK. 2010b. West Nile virus emergence and persistence in Los Angeles, California, 2003-2008. Am. J. Trop. Med. Hyg. [in press]

Lewis MA, Renclawowicz J, van den Driessche P, Wonham M. 2006. A comparison of continuous and discrete-time West Nile virus models. Bull. Math. Biol. 68:491-509.

Longshore J 1960. California encephalitis surveillance program. Introduction. Am. J. Hyg. 71:363-367.

Longshore WAJ, Stevens IM, Hollister Jr AC, Gittelsohn A, Lennette EH. 1956. Epidemiologic observations on acute infectious encephalitis in California, with special reference to the 1952 outbreak. Am. J. Hyg. 63:69-86.

Meyer KF, Haring CM, Howitt B. 1931. The etiology of epizootic encephalomyelitis of horses in the San Joaquin Valley, 1930. Science 74:227-228.

Moore CG, McLean RG, Mitchell CJ, Nasci RS, Tsai TF, Calisher CH, Marfin AA, Moore PS, Gubler DJ. 2002. Guidelines for arbovirus surveillance programs in the United States. Ft. Collins, CO: DVBID, CDC,PHS, US Dept Hlth Human Svcs.

Muckenfuss RS, Armstrong C, Webster LT. 1934. Etiology of the 1933 epidemic of encephalitis. J. A. M. A. 103:731-733.

Mulhern TD 1953. The use of mechanical traps in measuring mosquito populations. Proc. Calif. Mosq. Control Assoc. 21:64-66.

Newhouse VF, Chamberlain RW, Johnston Jr JG, Sudia WD. 1966. Use of dry ice to increase mosquito catches of the CDC miniature light trap. Mosq. News 26:30-35.

Patiris PJ, Oceguera LF, III, Peck GW, Chiles RE, Reisen WK, Hanson CV. 2008. Serologic diagnosis of West Nile and St. Louis encephalitis virus infections in domestic chickens. Am. J. Trop. Med. Hyg. 78:434-441.

Reeves WC 1985. The history of mosquito control in California. Proc. Mosq. Vector Control Assoc. Calif. 53:37-40.

Reeves WC, Asman SM, Hardy JL, Milby MM, Reisen WK. 1990. Epidemiology and control of mosquito-borne arboviruses in California, 1943-1987. Sacramento, Calif: Calif.Mosq.Vector Control Assoc.

Reeves WC, Hammon WM. 1962. Epidemiology of the arthropod-borne viral encephalitides in Kern County, California, 1943-1952. Univ. Calif. Berkeley Publ. Publ. Hlth. 4:1-257.

Reisen WK 1989. Estimation of vectorial capacity: relationship to disease transmission by malaria and arbovirus vectors. Bull. Soc. Vector Ecol. 14:39-40.

Reisen,WK. 1995. Guidelines for surveillance and control of arboviral encephalitis in California. 1-34.

Reisen WK, Barker CM. 2008. Use of climate variation in vectorborne disease decision support systems Global Climate Change and Extreme Weather Events: Understanding the Potential Contributions to the Emergence, Reemergence and Spread of Infectious Disease. Washington, DC.: Instute of Medicine. p 198-218.

Reisen WK, Cayan D, Tyree M, Barker CM, Eldridge BF, Dettinger M. 2008. Impact of climate variation on mosquito abundance in California. J. Soc. Vector Ecol. 33:89-98.

Reisen WK, Fang Y, Martinez VM. 2006. Effects of temperature on the transmission of West Nile virus by Culex tarsalis (Diptera: Culicidae). J Med. Entomol. 43:309-317.

Reisen WK, Lothrop HD, Chiles RE, Madon MB, Cossen C, Woods L, Husted S, Kramer VL, Edman JD. 2004. West Nile Virus in California. Emerg. Infect. Dis. 10:1369-1378.

Reisen WK, Presser SB, Lin J, Enge B, Hardy JL, Emmons RW. 1994. Viremia and serological responses in adult chickens infected with western equine encephalomyelitis and St. Louis encephalitis viruses. J. Am. Mosq. Control Assoc. 10:549-555.

Trawinski PR, Mackay DS. 2008. Meteorologically Conditioned Time-Series Predictions of West Nile Virus Vector Mosquitoes. Vector Borne. Zoonotic. Dis.

Walsh,JD. 1987. California’s mosquito encephalitis virus surveillance and control program.

Wegbreit J, Reisen WK. 2000. Relationships among weather, mosquito abundance and encephalitis virus activity in California: Kern County 1990-1998. J. Am. Mosq. Control Assoc. 16:22-27.

Wheeler SS, Barker CM, Armijos MV, Carroll BD, Husted SR, Reisen WK. 2009. Differential impact of West Nile virus on California birds. The Condor 111:1-20.

Wonham MJ, Camino-Beck T, Lewis MA. 2004. An epidemiological model for West Nile virus: invasion analysis and control applications. Proc. R. Soc. Lond B Biol. Sci. 271:501-507.