• View in gallery

    Locations of Larimer County sites (black circles and triangles) sampled for Cx. tarsalis in 2006 and of the Poudre River and Big Thompson River. Sites included in modeling of abundance of Cx. tarsalis based on climate data are shown as black circles. (Inset) Map showing locations of Larimer County (dark gray) and Boulder and Jefferson counties (light gray) in Colorado.

  • View in gallery

    Projected entomological risk of exposure to Cx. tarsalis in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range based on a cooling degree day model and where non-irrigated areas located more than 500 m from perceived larval habitat are included in the lowest abundance category. This figure appears in color at www.ajtmh.org.

  • View in gallery

    West Nile virus disease incidence per 100,000 person-years, 2002–2006, by census tract in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Census tract-based risk-classification index for combined entomological and epidemiological risk of exposure to Cx. tarsalis and West Nile virus in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range. This figure appears in color at www.ajtmh.org.

  • View in gallery

    West Nile virus disease incidence per 100,000 person-years, 2002–2006, in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range in relation to (A) percentage of census tract covered by area with high projected entomological risk of exposure to Cx. tarsalis and (B) distance to an area with high projected entomological risk (distance = 0 m if such an area is contained within a specific census tract). WNV disease incidences are based on combined case and population data for all census tracts falling within a given category for coverage of or distance to entomological risk area.

  • View in gallery

    Relationship between coverage of census tract by area with high projected entomological risk of exposure to Cx. tarsalis and West Nile virus disease incidence per 100,000 person-years, 2002–2006, in 5 different regions of Colorado: the western Colorado high plateau and mountains, the Larimer–Boulder–Jefferson 3-county portion of the northern Front Range (indicated by bold county outlines), the northern Front Range, the southern Front Range, and the eastern Colorado plains. WNV disease incidences are based on combined case and population data for all census tracts within each region falling within a given category for coverage of area with high entomological risk.

  • View in gallery

    Spatial distribution of areas with high projected entomological risk of exposure to Cx. tarsalis in the northern Front Range and western Colorado in relation to census tract-based West Nile virus disease incidence per 100,000 person-years, 2002–2006. This figure appears in color at www.ajtmh.org.

  • View in gallery

    Relationship between elevation (x) and mean daily temperature during June–August 2006 (y) along the Poudre River corridor in Larimer County (y = 28.64793 − 0.00538x; ANOVA; F1,6 = 22.76, r2 = 0.791, P = 0.003).

  • 1

    Goddard LB, Roth AE, Reisen WK, Scott TW, 2002. Vector competence of California mosquitoes for West Nile virus. Emerg Infect Dis 8 :1385–1391.

    • Search Google Scholar
    • Export Citation
  • 2

    Turell MJ, Dohm DJ, Sardelis MR, O’Guinn ML, Andreadis TG, Blow JA, 2005. An update on the potential of North American mosquitoes (Diptera: Culicidae) to transmit West Nile virus. J Med Entomol 42 :57–62.

    • Search Google Scholar
    • Export Citation
  • 3

    Turell MJ, Sardelis MR, O’Guinn ML, Dohm DJ, 2002. Potential vectors of West Nile virus in North America. Curr Top Microbiol Immunol 267 :241–252.

    • Search Google Scholar
    • Export Citation
  • 4

    Bolling BG, Moore CG, Anderson SL, Blair CD, Beaty BJ, 2007. Entomological studies along the Colorado Front Range during a period of intense West Nile virus activity. J Am Mosq Contr Assoc 23 :37–46.

    • Search Google Scholar
    • Export Citation
  • 5

    Smith GC, Moore CG, Davis T, Savage HM, Thapa AB, Shrestha SL, Karabatsos N, 1993. Arbovirus surveillance in northern Colorado, 1987 and 1991. J Med Entomol 30 :257–261.

    • Search Google Scholar
    • Export Citation
  • 6

    Gujral IB, Zielinski-Gutierrez EC, LeBailly A, Nasci R, 2007. Behavioral risks for West Nile virus disease, northern Colorado, 2003. Emerg Infect Dis 13 :419–425.

    • Search Google Scholar
    • Export Citation
  • 7

    Tempelis CH, Francy DB, Hayes RO, Lofy MF, 1967. Variations in feeding patterns of seven culicine mosquitoes on vertebrate hosts in Weld and Larimer counties, Colorado. Am J Trop Med Hyg 16 :111–119.

    • Search Google Scholar
    • Export Citation
  • 8

    Tsai TF, Smith GC, Happ CM, Kirk LJ, Jakob WL, Bolin RA, Francy DB, Lampert KJ, 1989. Surveillance of St. Louis encephalitis virus vectors in Grand Junction, Colorado, in 1987. J Am Mosq Contr Assoc 5 :161–165.

    • Search Google Scholar
    • Export Citation
  • 9

    Tsai TF, Smith GC, Ndukwu M, Jakob WL, Happ CM, Kirk LJ, Francy DB, Lampert KJ, 1988. Entomologic studies after a St. Louis encephalitis epidemic in Grand Junction, Colorado. Am J Epidemiol 128 :285–297.

    • Search Google Scholar
    • Export Citation
  • 10

    Reisen WK, Lothrop HD, 1995. Population ecology and dispersal of Culex tarsalis (Diptera: Culicidae) in the Coachella Valley of California. J Med Entomol 32 :490–502.

    • Search Google Scholar
    • Export Citation
  • 11

    Wegbreit J, Reisen WK, 2000. Relationships among weather, mosquito abundance, and encephalitis virus activity in California: Kern County 1990–98. J Am Mosq Contr Assoc 16 :22–27.

    • Search Google Scholar
    • Export Citation
  • 12

    Hess AD, Hayes RO, 1970. Relative potentials of domestic animals for zooprophylaxis against mosquito vectors of encephalitis. Am J Trop Med Hyg 19 :327–334.

    • Search Google Scholar
    • Export Citation
  • 13

    Lothrop HD, Reisen WK, 2001. Landscape affects the host-seeking patterns of Culex tarsalis (Diptera: Culicidae) in the Coachella Valley of California. J Med Entomol 38 :325–332.

    • Search Google Scholar
    • Export Citation
  • 14

    Baker M, 1961. The altitudinal distribution of mosquito larvae in the Colorado Front Range. Trans. Am. Entomol. Soc. 87 :231–246.

  • 15

    Harmston FC, Lawson FA, 1967. Mosquitoes of Colorado. Atlanta, GA: U.S. Department of Health, Education and Welfare, Public Health Service.

  • 16

    Carpenter SJ, 1969. Observations on the distribution and ecology of mountain Aedes mosquitoes in California. XII. Other species found in the mountains. Calif. Vector Views 16 :27–34.

    • Search Google Scholar
    • Export Citation
  • 17

    Darsie RF Jr, Ward RA, 2005. Identification and Geographical Distribution of the Mosquitoes of North America, North of Mexico. Gainesville, FL: University Press of Florida.

  • 18

    Bailey SM, Eliason DA, Hoffman BL, 1965. Flight and dispersal of the mosquito Culex tarsalis Coquillett in the Sacramento Valley of California. Hilgardia 37 :73–113.

    • Search Google Scholar
    • Export Citation
  • 19

    Dow RP, Reeves WC, Bellamy RE, 1965. Dispersal of female Culex tarsalis into a larvicided area. Am J Trop Med Hyg 14 :656–670.

  • 20

    Reisen WK, Milby MM, Meyer RP, Pfuntner AR, Spoehel J, Hazelrigg JE, Webb JP Jr, 1991. Mark–release–recapture studies with Culex mosquitoes (Diptera: Culicidae) in southern California. J Med Entomol 28 :357–371.

    • Search Google Scholar
    • Export Citation
  • 21

    Sall J, Creighton L, Lehman A, 2005. JMP Start Statistics. Third edition. Belmont, CA: Brooks/Cole.

  • 22

    Leung LR, Qian Y, Bian X, Washington WM, Han J, Roads JO, 2004. Mid-century ensemble regional climate change scenarios for the western United States. Clim Change 62 :75–113.

    • Search Google Scholar
    • Export Citation
  • 23

    Diuk-Wasser MA, Brown HE, Andreadis TG, Fish D, 2006. Modeling the spatial distribution of mosquito vectors for West Nile virus in Connecticut, USA. Vector Borne Zoonotic Dis 6 :283–295.

    • Search Google Scholar
    • Export Citation
  • 24

    Zou L, Miller SN, Schmidtmann ET, 2006. Mosquito larval habitat mapping using Remote Sensing and GIS: implications of coalbed methane development and West Nile virus. J Med Entomol 43 :1034–1041.

    • Search Google Scholar
    • Export Citation
  • 25

    Shone SM, Ferrao PN, Lesser CR, Norris DE, Glass GE, 2001. Analysis of mosquito vector species abundances in Maryland using geographic information systems. Ann NY Acad Sci 951 :364–368.

    • Search Google Scholar
    • Export Citation
  • 26

    Shaman J, Stieglitz M, Stark C, Le Blancq S, Cane M, 2002. Using a dynamic hydrology model to predict mosquito abundances in flood and swamp water. Emerg Infect Dis 8 :6–13.

    • Search Google Scholar
    • Export Citation
  • 27

    Rios J, Hacker CS, Hailey CA, Parsons RE, 2006. Demographic and spatial analysis of West Nile virus and St. Louis encephalitis in Houston, Texas. J Am Mosq Contr Assoc 22 :254–263.

    • Search Google Scholar
    • Export Citation
  • 28

    Kronenwetter-Koepel TA, Meece JK, Miller CA, Reed KD, 2005. Surveillance of above- and below-ground mosquito breeding habitats in a rural Midwestern community: baseline data for larvicidal control measures against West Nile virus vectors. Clin Med Res 3 :3–12.

    • Search Google Scholar
    • Export Citation
  • 29

    Gibbs SEJ, Wimberly MC, Madden M, Masour J, Yabsley MJ, Stallknecht DE, 2006. Factors affecting the geographic distribution of West Nile virus in Georgia, USA: 2002–2004. Vector Borne Zoonotic Dis 6 :73–82.

    • Search Google Scholar
    • Export Citation
  • 30

    Theophilides CN, Ahearn SC, Grady S, Merlino M, 2003. Identifying West Nile virus risk areas: the dynamic continuous-area space–time system. Am J Epidemiol 157 :843–854.

    • Search Google Scholar
    • Export Citation
  • 31

    Ward MP, 2006. Spread of equine West Nile virus encephalomyelitis during the 2002 Texas epidemic. Am J Trop Med Hyg 74 :1090–1095.

  • 32

    Ward MP, Ramsay BH, Gallo K, 2005. Rural cases of equine West Nile virus encephalomyelitis and the normalized difference vegetation index. Vector Borne Zoonotic Dis 5 :181–188.

    • Search Google Scholar
    • Export Citation
  • 33

    Brownstein JS, Rosen H, Purdy D, Miller JR, Merlino M, Mostashari F, Fish D, 2002. Spatial analysis of West Nile virus: rapid risk assessment of an introduced vector-borne zoonosis. Vector Borne Zoonotic Dis 2 :157–164.

    • Search Google Scholar
    • Export Citation
  • 34

    Corrigan RLA, Waldner C, Epp T, Wright J, Whitehead SM, Bangura H, Young E, Townsend HGG, 2006. Prediction of human cases of West Nile virus by equine cases, Saskatchewan, Canada, 2003. Prev Vet Med 76 :263–272.

    • Search Google Scholar
    • Export Citation
  • 35

    Yiannakoulias NW, Schopflocher DP, Svenson LW, 2006. Modelling geographic variations in West Nile virus. Can J Publ Health 97 :374–378.

  • 36

    Ruiz MO, Tedesco C, McTighe T, Austin C, Kitron U, 2004. Environmental and social determinants of human risk during a West Nile virus outbreak in the greater Chicago area, 2002. Int J Health Geogr 3 :8–19.

    • Search Google Scholar
    • Export Citation
  • 37

    Ruiz MO, Walker ED, Foster ES, Haramis LD, Kitron U, 2007. Association of West Nile virus illness and urban landscapes in Chicago and Detroit. Int J Health Geogr 6 :10–20.

    • Search Google Scholar
    • Export Citation
  • 38

    Ward MR, Stallknecht DE, Willis J, Conroy MJ, Davidson WR, 2006. Wild bird mortality and West Nile virus surveillance: biases associated with detection, reporting, and carcass persistence. J Wildl Dis 42 :92–106.

    • Search Google Scholar
    • Export Citation
  • 39

    Patnaik JL, Juliusson L, Vogt RL, 2007. Environmental predictors of human West Nile virus infections, Colorado. Emerg Infect Dis 13 :1788–1790.

    • Search Google Scholar
    • Export Citation
  • 40

    Eisen L, Eisen RJ, 2007. Need for improved methods to collect and present spatial epidemiologic data for vectorborne diseases. Emerg Infect Dis 13 :1816–1820.

    • Search Google Scholar
    • Export Citation
  • 41

    Eisen RJ, Eisen L, 2008. Spatial modeling of human risk of exposure to vector-borne pathogens based on epidemiological versus arthropod vector data. J Med Entomol 45 :181–192.

    • Search Google Scholar
    • Export Citation
  • 42

    Lothrop HD, Reisen WK, 1999. A geographical information system to manage mosquito and arbovirus surveillance and control data in the Coachella Valley of California. J Am Mosq Contr Assoc 15 :299–307.

    • Search Google Scholar
    • Export Citation
  • 43

    Takeda T, Whitehouse CA, Brewer M, Gettman AD, Mather TN, 2003. Arbovirus surveillance in Rhode Island: assessing potential ecologic and climatic correlates. J Am Mosq Contr Assoc 19 :179–189.

    • Search Google Scholar
    • Export Citation
  • 44

    Brownstein JS, Holford TR, Fish D, 2004. Enhancing West Nile virus surveillance, United States. Emerg Infect Dis 10 :1129–1133.

  • 45

    Cooke W III, Grala K, Wallis R, 2006. Avian GIS models signal human risk for West Nile virus in Mississippi. Int J Health Geogr 5 :36–56.

    • Search Google Scholar
    • Export Citation
  • 46

    Kutz FW, Wade TG, Pagac BB, 2003. A geospatial study of the potential of two exotic species of mosquitoes to impact the epidemiology of West Nile virus in Maryland. J Am Mosq Contr Assoc 19 :190–198.

    • Search Google Scholar
    • Export Citation
  • 47

    Lukacik G, Anand M, Shusas EJ, Howard JJ, Oliver J, Chen H, Backenson PB, Kauffman EB, Bernard KA, Kramer LD, White DJ, 2006. West Nile virus surveillance in mosquitoes in New York State, 2000–2004. J Am Mosq Contr Assoc 22 :264–271.

    • Search Google Scholar
    • Export Citation
  • 48

    Mongoh MN, Kaitsa ML, Dyer NW, 2006. Environmental and ecological determinants of West Nile virus occurrence in horses in North Dakota, 2002. Epidemiol Infect 135 :57–66.

    • Search Google Scholar
    • Export Citation
  • 49

    Orme-Zavaleta J, Jorgensen J, D’Ambrosio B, Altendorf E, Rossignol PA, 2006. Discovering spatio-temporal models of the spread of West Nile virus. Risk Anal 26 :413–422.

    • Search Google Scholar
    • Export Citation
  • 50

    Eisen RJ, Lane RS, Fritz CL, Eisen L, 2006. Spatial patterns of Lyme disease risk in California based on disease incidence data and modeling of vector-tick exposure. Am J Trop Med Hyg 75 :669–676.

    • Search Google Scholar
    • Export Citation

 

 

 

 

Combining Mosquito Vector and Human Disease Data for Improved Assessment of Spatial West Nile Virus Disease Risk

View More View Less
  • 1 Department of Microbiology, Immunology, and Pathology, Colorado State University, Fort Collins, Colorado; Division of Vector-Borne Infectious Diseases, National Center for Vector-Borne, Zoonotic and Enteric Diseases, Centers for Disease Control and Prevention, Fort Collins, Colorado; Colorado Department of Public Health and Environment, Denver, Colorado

Assessments of spatial risk of exposure to vector-borne pathogens that combine vector and human disease data are needed for areas encompassing large tracts of public land with low population bases. We addressed this need for West Nile virus (WNV) disease in the northern Colorado Front Range by developing not only a spatial model for entomological risk of exposure to Culex tarsalis WNV vectors and an epidemiological risk map for WNV disease but also a novel risk-classification index combining data for these independently derived measures of entomological and epidemiological risk. Risk of vector exposure was high in the densely populated eastern plains portion of the Front Range but low in cooler montane areas to the west that are sparsely populated but used heavily for recreation in the summer. The entomological risk model performed well when applied to the western, mountainous part of Colorado and validated against epidemiologic data.

INTRODUCTION

West Nile virus (WNV) disease has emerged as a serious problem in the Central Plains and Rocky Mountains with epidemics causing hundreds to thousands of reported human cases in Colorado, Montana, Nebraska, Wyoming, and the Dakotas in 2003 and in Idaho and Utah in 2006 (data from the Centers for Disease Control and Prevention; http://www.cdc.gov/ncidod/dvbid/westnile). In 2003, Colorado alone reported 2,947 cases of human disease with 63 fatalities (data from the Colorado Department of Public Health and Environment; http://www.cdphe.state.co.us/dc/zoonosis/wnv/). The northern part of the Colorado Front Range, which spans the transition zone from the Central Plains to the Rocky Mountains, was one of the focal points of this epidemic with 3 counties (Larimer, Boulder, Jefferson) accounting for 1,124 reported WNV cases in 2003 and 187 additional cases during 2004–2006. The caseload for this 3-county area notably exceeds that of most eastern states.

Spatial patterns of risk of human exposure along the northern Colorado Front Range to locally occurring WNV vectors such as Culex pipiens and, especially, Culex tarsalis17 are intriguing because this area 1) extends from the western edge of the Central Plains into the eastern edge of the Rocky Mountains and encompasses habitats ranging from prairie through foothills shrub, montane and subalpine forest, and to alpine areas; 2) includes dramatic elevation–climate gradients potentially affecting the ability of Culex mosquitoes to establish breeding populations; and 3) exhibits strong spatial clustering of the human population to the plains habitat in the east.

Cx. tarsalis, because of its high vector efficiency,1,2 common occurrence,46,811 and willingness to bite mammals,7,12,13 is considered a primary vector of WNV to humans in Colorado and indeed the western United States. Along the northern Colorado Front Range, this mosquito occurs commonly in heavily populated plains areas at low elevations (e.g., the cities of Boulder, Loveland, and Fort Collins).46 There is, however, a lack of detailed knowledge regarding the distribution and abundance of Cx. tarsalis in montane areas to the west. These montane areas include large tracts of public land (e.g., Rocky Mountain National Park, Roosevelt National Forest) used heavily in the summer for recreational activities by the Front Range population as well as tourists. Information regarding Cx. tarsalis in montane areas of the western United States typically has been qualitative and restricted to records of collection sites.1416 Because montane areas in the northern Colorado Front Range and, indeed, many other parts of the western United States are dominated by public-access lands and have very low population bases, epidemiologic data are of limited use to assess risk of WNV exposure and need to be complemented with models for entomological risk of exposure to Cx. tarsalis and other potential WNV vectors.

The primary goal of this study was to create a comprehensive spatial risk model, combining entomological risk of exposure to Cx. tarsalis and epidemiological risk of WNV exposure, for a 3-county area in the northern Colorado Front Range extending from prairie landscapes at the western edge of the Central Plains into montane and subalpine/alpine habitats at the eastern edge of the Rocky Mountains. A secondary goal was to explore how the entomological risk model performed, compared with epidemiologic data, when scaled-up from the ecologically and climatically diverse Larimer County model development area to 4 different regions of Colorado: eastern plains, northern Colorado Front Range, southern Colorado Front Range, and western mountains and high plateau.

MATERIALS AND METHODS

Development of entomological risk model: field sampling of Cx. tarsalis.

Fieldwork targeted 2 habitat-climate-elevation gradients in the central and southern parts of Larimer County: the Poudre River and Big Thompson River corridors (Figure 1). The general climate in this area is characterized by cold winters and hot summers with low humidity, averaging ~ 400 mm of precipitation per year at lower elevations (Mountain States Weather Services, Fort Collins, CO). Mosquito sampling was conducted on 5–7 occasions (roughly every 2 weeks in each site) during June 20–September 13, 2006. This time period includes the peak seasonal activity by Cx. tarsalis in Larimer County.4,5 Sampling included 10 sites located along each river corridor (total of 20 sites) at elevations ranging from ~ 1,500 to 2,350 m (Figure 1). Mosquitoes were collected using CO2-baited CDC miniature light traps (John W. Hock Company, Gainesville, FL) suspended ~ 1.5 m above the ground and operated from afternoon (1500–1700 hours) until morning (0800–1000 hours). Each sampling site held 2 traps baited with ~ 1 kg of dry ice and located directly along the aforementioned rivers or their tributaries. Trees lining the sampling sites included cottonwood (Populus spp.), willow (Salix spp.), and, at higher elevations, aspen (Populus tremuloides). For sites located along the Poudre River, we also determined temperature and relative humidity, at ~ 1 m above the ground on the north-facing side of a tree, every 30 min throughout the study period using HOBO H8 Pro series loggers (Onset Computer Corporation, Pocasset, MA).

Collected mosquitoes were examined with a dissecting microscope and identified to species using published keys.15,17 Site-specific mean abundances of Cx. tarsalis per trap-night were calculated for the period June 20–September 13, 2006; this included data from 10–14 trap-nights in each site. Mean abundance over a prolonged time period was used to minimize the effects of 1) variable weather conditions on different trapping nights on mosquito activity and 2) differences in the seasonal timing of peak abundance of Cx. tarsalis along the elevation–climate gradients examined.

Development of entomological risk model: Geographic Information System database.

Site locations were mapped with a GPS receiver (Trimble Geo XT; Trimble Corp., Sunnyvale, CA). Site-specific 30-yr mean climate data (mean monthly and annual minimum, mean, and maximum temperature; mean monthly and annual cooling, heating, and growing degree days; mean monthly and annual precipitation, snowfall, and relative humidity; mean and median annual length of freeze-free period; median Julian date of first and last snowfall) were derived from Geographic Information System (GIS)-based data for 1961–1990 (2 × 2 km spatial resolution; Climate Source LLC, Corvallis, OR) using ArcGIS9.2 (ESRI, Redlands, CA). Climate data layers were generated by Climate Source LLC using the PRISM modeling system and Gaussian filter resolution enhancement. Site-specific data on elevation were derived from the U.S. Geological Survey 30-m digital national elevation dataset. Selected environmental site characteristics are provided in Table 1.

Development of entomological risk model: model construction.

Associations between site-specific long-term climate data (annual and monthly data for 1961–1990) and mean abundance of Cx. tarsalis per trap-night during June 20–September 13, 2006 (hereafter referred to as abundance of Cx. tarsalis) were explored by multivariate regression models based on data from 13 sites yielding at least 0.2 mosquito per trap-night. This excluded 7 sites located at elevations above 1,950 m and found to yield no or very few mosquitoes (0–0.1 per trap-night). Long-term climate data are useful for recognizing relative differences in climatic conditions between spatial locations and thus can serve as the basis for spatial modeling. Our use of climate data from 1961 to 1990 and mosquito data from 2006 was based on the assumption that differences in climatic conditions between sampling sites in 2006 were consistent with the differences for the long-term averages. This indeed was the case for one key climate factor available both as temperature logger-derived data from 2006 and long-term 1961–1990 GIS-derived data; site-specific mean June–August temperatures in 2006 were strongly correlated with 1961–1990 averages for these months in the same sites (pairwise correlation coefficient = 0.891, N = 8, P = 0.003). Use of long-term climate data available as spatially continuous GIS data layers also enabled development of spatially continuous predictive models for mosquito abundance based on a far wider range of potentially important climate factors than those available for the mosquito sampling year of 2006.

Elevation, climate factors, and mosquito abundance (natural log-transformed data) were all normally distributed (Shapiro–Wilk test; P > 0.05 in all cases). Results for selected environmental variables in univariate tests are shown in Table 2. A model to predict abundance of Cx. tarsalis (dependent variable) was constructed using a forward stepwise multiple regression approach including five biologically meaningful covariates (Table 2; mean temperature for April–September; cumulative cooling degree days for June–August; mean relative humidity for March–May; mean precipitation for January–February; snowfall in April) and using a probability to enter covariates of 0.25.

The equation for the resulting model (which was based on the relationship between cooling degree days and abundance of Cx. tarsalis) was applied, at a 2 × 2 km spatial resolution, to the cooling degree day data layer using the raster calculator of ArcGIS9.2 to create a continuous data layer predictive of abundance of Cx. tarsalis. Inclusion of Boulder and Jefferson counties in this model is justified because ranges for the single covariate included in the model (cooling degree days) are similar in these counties to the range occurring in the Larimer County model development area; the minimum cooling degree day value was identical for all 3 counties, and the maximum values for Boulder and Jefferson counties exceeded that for Larimer County by no more than 20% (Table 3). Further, the areas of Boulder and Jefferson counties with cooling degree day values exceeding the maximum for Larimer County were only 17% and 14% of the total areas of these counties, respectively.

The model for projected entomological risk of exposure to Cx. tarsalis was categorized, at a spatial resolution of 2 × 2 km, as follows: low (< 1 mosquito per trap-night), moderate (1–5 mosquitoes per trap-night), or high (> 5 mosquitoes per trap-night). This cooling degree day-based model for abundance of Cx. tarsalis does not, however, account for presence of open, dry habitats with low suitability for Cx. tarsalis within areas classified as moderate or high risk. We therefore created an analysis mask based on GIS data layers for presence of surface water (streams, rivers, irrigation canals, lakes, ponds) obtained from the U.S. Geological Survey National Hydrography Dataset and landcover from the Colorado Gap Analysis Project (CO-GAP). The mask was designed to uniformly include nonirrigated areas (natural habitats or nonirrigated cattle grazing land) at distances exceeding a given cut-off value (hereafter referred to as buffer distance) from perceived larval habitat into the lowest category of entomological risk (< 1 mosquito per trap-night). Cx. tarsalis can disperse over large distances in irrigated agricultural settings in California,10,1820 but empirical data for dispersal of adults from larval habitats in the semiarid landscapes of Colorado have been lacking. We initially considered 4 different potential buffer distances: 100, 250, 500, and 1,000 m. However, the 100- and 250-m buffer distances proved unrealistic as preliminary data collected from 3 separate sites in southeastern Larimer County (Butterfly Woods Natural Area, Dixon Reservoir, Douglass Reservoir) during June 28–July 12, 2007, indicate that abundance of Cx. tarsalis can remain high (> 5 mosquitoes per trap-night) at distances up to 250 m from larval habitat detectable from GIS-based data layers (data not shown). Spatial models predicting moderate or high entomological risk of exposure to Cx. tarsalis were therefore developed using a conservative buffer distance of 500 m around perceived larval habitat (streams, rivers, irrigation canals, lakes, ponds, irrigated agricultural land).

Development of epidemiological risk maps.

The epidemiological risk map for the Larimer–Boulder–Jefferson area was based on data for WNV disease cases reported to the Colorado Department of Public Health and Environment from 2002 to 2006 and with a known census tract of residence; this included 618 cases for Larimer County, 514 cases for Boulder County, and 181 cases for Jefferson County. Census tract population data used were from 2004 (data provided by ESRI and based on information from the U.S. Census Bureau). Cumulative incidence (hereafter referred to as incidence) of WNV disease per 100,000 person-years, 2002–2006, was calculated and displayed by census tract using ArcGIS9.2. Epidemiological risk similarly was mapped for the entire state of Colorado; this included data for 3,749 WNV disease cases occurring during 2002–2006 and with a known census tract of residence.

Comparison of entomological and epidemiological risk.

The Spatial Analyst in ArcGIS9.2 was used to calculate the percentage of each census tract in Larimer–Boulder–Jefferson covered by area representing moderate (> 1 Cx. tarsalis per trap-night) or high (> 5 Cx. tarsalis per trap-night) projected entomological risk of exposure to Cx. tarsalis. We also created a 100 × 100 m grid and calculated average straight line (Euclidean) distances from the center of a census tract to the nearest area with moderate or high projected entomological risk (distance = 0 m if such areas were contained within the census tract). Thereafter, associations were explored between percentage of a census tract classified as area with moderate or high projected entomological risk or average distance to the nearest area with moderate or high projected entomological risk and WNV disease incidence. Because results for moderate and high entomological risk were similar, only data for comparisons between high entomological risk and WNV disease incidence are presented here.

The potential for applying the entomological risk model beyond the Larimer–Boulder–Jefferson area was evaluated by determination of the strength of the association between percentage of census tract classified as area with high projected entomological risk and WNV disease incidence for four different regions of Colorado: the eastern plains, the northern Colorado Front Range, the southern Colorado Front Range, and the western mountains and high plateau (see Figure 6 for a map showing the area included in each region). Statistical analyses were carried out using the JMP statistical package21 and results were considered significant when P < 0.05.

RESULTS

Entomological risk model for the Larimer–Boulder–Jefferson area.

We used a forward stepwise regression modeling approach to develop a model capable of explaining variability in abundance of Cx. tarsalis (Table 1) from climate data with relevance to mosquito biology and available as GIS data layers at a 2 × 2 km spatial resolution (Table 2; mean temperature for April–September; cumulative cooling degree days for June–August; mean relative humidity for March–May; mean precipitation for January–February; snowfall in April). The resulting model was based on cumulative cooling degree days for June–August (x) (ln Cx. tarsalis per trap-night = −3.2485 + 0.0122x; ANOVA: F1,11 = 28.22, adjusted r2 = 0.758, P < 0.001). Lack-of-fit test indicated that a sufficient number of independent variables (N = 1) were included in the model (P = 0.41). There was no spatial autocorrelation either for abundance of Cx. tarsalis in the sites included in the modeling effort (Moran’s I = 0.01, Z[I] = 1.1) or for the residuals of the linear regression of abundance of Cx. tarsalis on cooling degree days (I = −0.08, Z[I] = 0).

The equation for the model predicting abundance of Cx. tarsalis was applied to GIS-based cooling degree day data clipped to the Larimer–Boulder–Jefferson area and projected entomological risk of exposure to Cx. tarsalis was categorized as low (< 1 mosquito per trap-night), moderate (1–5 mosquitoes per trap-night), or high (> 5 mosquitoes per trap-night). Buffers around perceived larval habitat were applied to uniformly include nonirrigated areas located more than 500 m from larval habitat in the lowest risk category. The spatial pattern of areas with high projected entomological risk of exposure to Cx. tarsalis in the Larimer–Boulder–Jefferson area is shown in Figure 2. High-risk areas include southeastern Larimer County, eastern Boulder County, and northeastern Jefferson County.

Epidemiological risk map for the Larimer–Boulder–Jefferson area.

The epidemiological risk map for the Larimer–Boulder–Jefferson area is shown in Figure 3 with 5 risk categories: 0–10, 11–20, 21–40, 41–60, and > 60 WNV disease cases per 100,000 person-years from 2002 to 2006. Census tracts with WNV disease incidences > 60 cases per 100,000 person-years occurred exclusively in the eastern part of the targeted counties, especially from southeastern Larimer County through eastern Boulder County and into the northeastern edge of Jefferson County. Overall, WNV disease incidence decreased from the plains in eastern Larimer and Boulder counties (typically > 40 WNV cases per 100,000 person-years) to montane areas to the west in these counties or to Jefferson County in the south (most commonly < 20 WNV cases per 100,000 person-years).

Combined entomological and epidemiological risk-classification index for the Larimer–Boulder–Jefferson area.

We also created a census tract-based risk-classification index for combined entomological and epidemiological risk of exposure to vectors and WNV in the Larimer–Boulder–Jefferson area. This novel risk-classification index is based on percentage coverage by area with high projected risk of exposure to Cx. tarsalis and WNV disease incidence and includes 5 risk classes: Very Low, Low, Moderate, High, and Very High (see Table 4 for further description of these risk classes). The risk-classification index map (see Figure 4) indicates that areas with high to very high risk-classification indices cluster from southeastern Larimer County through Boulder County and into northeastern Jefferson County. The western, montane parts of Larimer and Boulder counties and most of Jefferson County were classified as low to very low risk.

Comparison of spatial patterns of entomological and epidemiological risk in the Larimer–Boulder–Jefferson area.

We found a strong association between presence of areas with high projected entomological risk of exposure to Cx. tarsalis and incidence of WNV disease at the census tract scale in the Larimer–Boulder–Jefferson area. WNV disease incidence was positively associated with percentage coverage (0, 1–33%, 34–67%, 68–100%) by areas with high projected entomological risk (Figure 5A; WNV disease incidences based on combined case and population data for all census tracts falling within a given category of entomological risk). Statistical analysis was based on data from individual census tracts and revealed a significant positive correlation between coverage of area by high projected entomological risk and WNV disease incidence (Spearman’s rank correlation; ρs = 0.409, N = 253, P < 0.001). Further, census tracts containing 68–100% area with high projected entomological risk had higher WNV disease incidence among residents (median of 26.9 cases per 100,000 person-years; N = 132 census tracts) than census tracts either lacking such areas (median of 4.7 cases per 100,000 person-years; N = 41 census tracts) or with 1–33% coverage (median of 11.4 cases per 100,000 person-years; N = 38 census tracts) or 34–67% coverage (median of 10.1 cases per 100,000 person-years; N = 42 census tracts) (Wilcoxon rank sum test with χ2 approximation: χ2 ≥ 10.68, df = 1, P < 0.002 in all three cases). Finally, WNV disease incidence was higher for census tracts with 34–67% or 1–33% coverage of areas with high projected entomological risk than for census tracts lacking such areas (χ2 ≥ 6.73, df = 1, P < 0.01 in both cases).

We also found that WNV disease incidence was negatively associated with average distance to nearest area with high projected entomological risk (Figure 5B; WNV disease incidences based on combined case and population data for all census tracts falling within a given distance category). Statistical analysis was again based on data from individual census tracts and revealed that WNV disease incidence for census tracts containing areas with high projected entomological risk (median of 17.7 cases per 100,000 person-years; N = 212 census tracts) was 4-fold higher than for census tracts lacking such areas (median of 4.7 cases per 100,000 person-years; N = 41 census tracts) (χ2 = 28.60, df = 1, P < 0.001).

Performance of entomological risk model, relative to epidemiologic data, when applied to different regions of Colorado.

We also determined how the entomological risk model performed, compared with epidemiologic data, when scaled-up from the ecologically and climatically diverse Larimer County model development area to 4 different regions of Colorado: eastern plains, northern Front Range, southern Front Range, and western mountains and high plateau (Figure 6). Overall, the entomological risk model performed well in the northern Front Range and western mountains and high plateau but very poorly in the eastern plains (Figure 6; WNV disease incidences based on combined case and population data for all census tracts within a region of the state falling within a given category of entomological risk). The percentage of an individual census tract covered by area classified as high projected entomological risk was significantly positively correlated with WNV disease incidence not only for the Larimer–Boulder–Jefferson area (ρs = 0.551, N = 253, P < 0.001) but also for an expanded portion of the northern Front Range (ρs = 0.356, N = 427, P < 0.001), western Colorado (ρs = 0.740, N = 151, P < 0.001), and the southern Front Range (ρs = 0.292, N = 75, P = 0.01). In contrast, we found a negative correlation between coverage with high projected entomological risk and WNV disease incidence in the eastern plains (ρs = −0.304, N = 422, P < 0.001).

Census tracts in the northern Front Range containing 68–100% area with high projected entomological risk had higher WNV disease incidence among residents (median of 10.5 cases per 100,000 person-years; N = 305 census tracts) than census tracts lacking such areas (median of 0 cases per 100,000 person-years; N = 77 census tracts) or with 1–33% coverage of area with high entomological risk (median of 4.7 cases per 100,000 person-years; N = 25 census tracts) (χ2 = 50.34, df = 1, P < 0.001 and χ2 = 5.69, df = 1, P = 0.02, respectively). A similar pattern was seen for western Colorado where census tracts containing 68–100% area with high projected entomological risk had higher WNV disease incidence among residents (median of 25.8 cases per 100,000 person-years; N = 32 census tracts) than census tracts lacking such areas (median of 0 cases per 100,000 person-years; N = 93 census tracts) or with 1–33% coverage of area with high entomological risk (median of 11.8 cases per 100,000 person-years; N = 17 census tracts) (χ2 = 76.95, df = 1, P < 0.001 and χ2 = 8.16, df = 1, P = 0.004, respectively). In the southern Front Range, census tracts containing 68–100% area with high projected entomological risk had higher WNV disease incidence among residents (median of 25.7 cases per 100,000 person-years; N = 63 census tracts) than census tracts with 0–33% coverage (median of 9.3 cases per 100,000 person-years; N = 7 census tracts) (χ2 = 6.63, df = 1, P = 0.01). In striking contrast, census tracts in eastern Colorado containing 100% area with high projected entomological risk had 6-fold lower WNV disease incidence among residents (median of 6.3 cases per 100,000 person-years; N = 391 census tracts) than census tracts with 0–67% coverage (median of 39.9 cases per 100,000 person-years; N = 20 census tracts) (χ2 = 24.08, df = 1, P < 0.001).

Figure 7 shows the scaled-up spatial model for high projected entomological risk of exposure to Cx. tarsalis, in relation to census tract-based WNV disease incidence, for the northern Front Range and western Colorado.

Potential impact of climate warming on mosquito abundance along climate–elevation gradients in the northern Front Range.

We used data from HOBO loggers operated in 8 sites along the Poudre River during 2006 to estimate the relationship between elevation (x) and daily mean temperature during the peak mosquito breeding season in June–August (Figure 8). Based on the linear relationship between daily mean June–August temperature in 2006 (y) and elevation (x) (y = 28.64793 − 0.00538x; F1,6 = 22.76, r2 = 0.791, P = 0.003), incremental elevation increases of 100 m along the Poudre River corresponded to an incremental decrease in mean daily June–August temperature in 2006 of 0.5°C per 100-m elevation change unit. It therefore follows that a realistic future climate-warming scenario resulting in mean daily summer temperature increases of 1–2°C22 would cause a shift toward current temperature conditions in the future occurring at elevations 200–400 m higher than today. This realistic scenario would most likely result in both expansion in the range of Cx. tarsalis toward higher elevations in the Colorado Front Range and increased mosquito abundance near the current upper altitudinal (climate) limit where Cx. tarsalis now is present but scarce.

DISCUSSION

Spatial assessments of risk of exposure to WNV have tended to focus on either entomological risk of vector exposure2328 or disease risk based on avian or equine WNV surveillance2932 or human case data.3337 However, each of these risk measures has inherent weaknesses. Entomological risk does not account for the importance of human behavior such as use of mosquito repellents.6 Avian WNV surveillance either requires costly serosurveys or is subject to variability in levels of detection of dead birds.38,39 Epidemiologic data from humans only provide reliable information for risk in areas with an adequate population base and typically are based on address of residence rather than likely exposure site.40,41 Following earlier efforts for other mosquito-borne arboviruses,42 several studies have attempted to address these issues by combining entomological risk measures (vector larval habitat, vector abundance, abundance of infected mosquito pools) with either avian WNV surveillance data43 or human WNV disease data (case locations, disease incidence).4449 However, these studies typically either were restricted to risk assessments at a crude county spatial scale44,47,48 or failed to generate continuous spatial risk surfaces.43,46,49 Perhaps the most complete previous approach comes from a Mississippi study combining spatial continuous environmental data with zip-code–based incidence of WNV disease in humans.45 This approach was, however, to some extent impeded by a low case load of human WNV disease (276 cases were reported from Mississippi during the 2002–2003 study period).

To account for the inherent weakness of using either entomological or epidemiological risk measures separately, we developed not only a spatial model for entomological risk of exposure to Cx. tarsalis (Figure 2) and an epidemiological risk map for WNV disease (Figure 3) but also a novel risk-classification index combining data for the independently derived measures of entomological and epidemiological risk (Figure 4). This included a 3-county area (Larimer–Boulder–Jefferson) in the northern Colorado Front Range severely afflicted by WNV disease; the 3 counties reported 1,313 human cases from 2002 to 2006. Risk of exposure to Cx. tarsalis and WNV was found to be high in the densely populated eastern plains portion of the northern Front Range but low in cooler montane areas to the west that are sparsely populated but used heavily for recreation.

Development of spatial risk models for vector-borne diseases in the western United States that combine entomological and epidemiological risk measures, as previously accomplished for Lyme disease in California,41,50 is crucial because this part of the country encompasses large tracts of public land where risk cannot be assessed by epidemiologic data alone due to low population bases. Public lands can, however, be heavily used for recreational purposes and thus represent significant risk of human exposure to vector-borne pathogens. In this study, we were able to use an entomological risk model to demonstrate that Rocky Mountain National Park in far southwestern Larimer County, which receives > 3 million annual visitors with a summer peak in visitor numbers coinciding with vector mosquito and WNV activity in Colorado,4,5 currently presents minimal risk of exposure to Cx. tarsalis vectors. This was corroborated by the fact that mosquito sampling during the summer of 2006 in 6 locations in a heavily used portion of Rocky Mountain National Park yielded a single Cx. tarsalis (L. Eisen, unpublished data).

Our epidemiological risk assessment highlights the importance of using an appropriate spatial scale, such as census tract or zip code, for presentation of incidence of vector-borne diseases in the west, where counties tend to cover large areas and often include considerable ecological and climatic variability.40 This is important both to detect areas of low risk within counties with overall high risk (e.g., WNV disease in the Larimer–Boulder–Jefferson area [Figure 3] or Lyme disease in north coastal California50) and to detect small, isolated areas of high risk within counties with overall low risk (e.g., WNV disease in parts of western Colorado [Figure 7] or Lyme disease in southern California50). Additional research efforts relating to spatial ecoepidemiology of WNV disease in Colorado are underway.

Access to census tract-based WNV disease incidence data from 2002 to 2006 for the state of Colorado provided an opportunity to examine how the entomological risk model for the Larimer–Boulder–Jefferson area performed when scaled-up to a larger area of the northern Front Range (Larimer–Boulder–Jefferson–Douglas–El Paso), the southern Front Range, the western Colorado high plateau and mountains, and the eastern Colorado plains. The robust performance, relative to epidemiologic data, of the scaled-up entomological risk model in western Colorado can be attributed to the Larimer County model development area including a climate gradient similar to that for western Colorado. This demonstrates the value of developing spatial models for entomological risk of vector exposure within small but topographically, climatically, and ecologically diverse geographical areas. On the other hand, the Larimer County-derived entomological risk model performed very poorly in the eastern plains, likely because Larimer County includes only a limited portion of the climate variability existing in eastern Colorado. Efforts are underway to generate data for abundance of Cx. tarsalis along a climate gradient extending from the Front Range into the eastern plains along the Big Thompson and South Platte rivers. These data will be used to develop a separate model for entomological risk of exposure to Cx. tarsalis in the eastern plains.

Finally, our data on abundance of Cx. tarsalis along climate–elevation gradients in the Colorado Front Range and the relationship between elevation and temperature in this area suggest that the spatial distribution and abundance patterns of Cx. tarsalis are sensitive to climate warming. We expect projected climate warming over the next 50 yr in Colorado, which includes summer temperature increases of 1–2°C,22 to shift the cool end of the distribution of Cx. tarsalis several hundred meters upward in elevation in the northern Front Range. This, in conjunction with potential changes in availability of mosquito larval habitat following expected decreases in mountain snowpack and subsequent river and stream flooding activity, will impact future patterns of risk of exposure to Cx. tarsalis in the Rocky Mountain region. The Colorado Front Range is exceptionally well suited for long-term empirical studies on the effect of climate warming on spatial patterns of distribution and abundance of Cx. tarsalis and spatial patterns of presence of WNV in local mosquito populations.

Table 1

Abundance of Cx. tarsalis from late June to mid-September 2006 and selected environmental characteristics for 20 sites in Larimer County, Colorado

Mean annual climate data*
SiteMean no. of Cx. tarsalis per trap-nightElevation (m)Temp. (°C)Cooling degree days†Relative humidity (%)Precipitation (mm)Snowfall (mm)
* Mean values for 1961–1990 based on GIS-derived data (2 × 2 km spatial resolution); site locations determined with a GPS receiver.
† Cooling degree days are calculated as the number of degree days exceeding a baseline of 65°F (18.3°C).
‡ Sites included in climate-based model for abundance of Cx. tarsalis.
Lee Martinez‡7.6 ± 11.51,5109.4495523921,396
Namaqua‡7.3 ± 4.31,5209.3479523661,155
Glade Park‡4.3 ± 3.71,5409.2459523891,147
Bellvue‡30.3 ± 58.91,5608.7450524141,445
Picnic Rock‡2.5 ± 3.31,6107.6381534591,843
Gateway‡0.4 ± 0.61,6407.7340534431,727
Narrows‡4.5 ± 7.21,6908.5388534361,303
Viestenz Smith‡4.6 ± 9.21,7307.8357534561,551
Ouzel‡0.2 ± 0.41,7506.5221544922,244
Idylwilde‡0.4 ± 0.81,8407.3280544621,679
Stove Prairie‡0.3 ± 0.81,8606.0174544962,347
Drake‡0.2 ± 0.41,8706.8280544671,778
Waltonia01,9706.7253544601,762
Dutch George‡0.2 ± 0.62,0005.7153544512,188
Big Thompson02,0906.4208544571,764
Eggers0.1 ± 0.32,1105.296533811,860
Sleepy Hollow02,1206.2155544561,743
Dadd Gulch0.1 ± 0.32,1304.892533301,853
Glen Comfort02,2205.974543931,199
Bliss0.1 ± 0.32,3601.90544653,418
Table 2

Adjusted r2 values and statistical significances for linear relationships between monthly or annual long-term (1961–1990) average climate data and abundance of Cx. tarsalis for 13 sites in Larimer County, Colorado, from late June to mid-September 2006

Adjusted r2 value for linear relationship between climate variable and abundance of Cx. tarsalis (natural log-transformed)†
Time periodMean temperatureCooling degree days ‡Relative humidityPrecipitationSnowfall
Climate variable-specific data shown in bold were used (as mean or cumulative data for indicated months) in a multivariate forward stepwise regression modeling approach to determine the association between climate conditions and abundance of Cx. tarsalis.
† ANOVA: NS, no significant association (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001.
‡ Cooling degree days are calculated as the number of degree days exceeding a baseline of 65°F (18.3°C).
January0.499**0NS0.047NS0.731***0.541**
February0.710***0NS0.520**0.715***0.532**
March0.708***0NS0.687***0.420**0.491**
April0.733***0NS0.615***0.333*0.665***
May0.719***0.300*0.670***0.051NS0.543**
June0.715***0.756***−0.046NS−0.083NS0NS
July0.717***0.754***0.470**0.201NS0NS
August0.712***0.759***0.529**0.496**0NS
September0.715***0.726***0.516**0.278*0.373*
October0.683***0.153NS−0.091NS0.369*0.553**
November0.679***0NS0.111NS0.687***0.461**
December0.590**0NS0.010NS0.612**0.541**
Annual0.715***0.755***0.796***0.498**0.579***
Table 3

Climate ranges for Larimer, Boulder, and Jefferson counties

Annual climate ranges (2 × 2 km spatial resolution)*
CountyMean temperature (°C)Cooling degree days†Relative humidity (%)Precipitation (mm)Snowfall (mm)
* Mean values for 1961–1990 based on GIS-derived data.
† Cooling degree days are calculated as the number of degree days exceeding a baseline of 65°F (18.3°C).
Larimer−0.1 to 9.70–54851–57321–1,265854–17,382
Boulder−0.1 to 10.30–66151–60355–1,255993–17,721
Jefferson3.7–10.00–63949–55364–6051,336–4,281
Table 4

Risk-classification index scheme for West Nile virus exposure in Larimer–Boulder–Jefferson based on percentage coverage of census tract by area with high projected risk of exposure to Cx. tarsalis and West Nile virus disease incidence

Risk-classification indexNo. of census tractsCoverage of census tract by area with high projected risk of exposure to Cx. tarsalisCensus tract-based WNV disease incidence per 100,000 person-years, 2002–2006
Very low290–10%< 5
Low270–10%> 5
Low1511–50%< 5
Moderate2811–50%> 5
Moderate71> 50%< 20
High41> 50%> 20
Very high42> 75%> 40
Figure 1.
Figure 1.

Locations of Larimer County sites (black circles and triangles) sampled for Cx. tarsalis in 2006 and of the Poudre River and Big Thompson River. Sites included in modeling of abundance of Cx. tarsalis based on climate data are shown as black circles. (Inset) Map showing locations of Larimer County (dark gray) and Boulder and Jefferson counties (light gray) in Colorado.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 2.
Figure 2.

Projected entomological risk of exposure to Cx. tarsalis in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range based on a cooling degree day model and where non-irrigated areas located more than 500 m from perceived larval habitat are included in the lowest abundance category. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 3.
Figure 3.

West Nile virus disease incidence per 100,000 person-years, 2002–2006, by census tract in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 4.
Figure 4.

Census tract-based risk-classification index for combined entomological and epidemiological risk of exposure to Cx. tarsalis and West Nile virus in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 5.
Figure 5.

West Nile virus disease incidence per 100,000 person-years, 2002–2006, in the Larimer–Boulder–Jefferson area of the northern Colorado Front Range in relation to (A) percentage of census tract covered by area with high projected entomological risk of exposure to Cx. tarsalis and (B) distance to an area with high projected entomological risk (distance = 0 m if such an area is contained within a specific census tract). WNV disease incidences are based on combined case and population data for all census tracts falling within a given category for coverage of or distance to entomological risk area.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 6.
Figure 6.

Relationship between coverage of census tract by area with high projected entomological risk of exposure to Cx. tarsalis and West Nile virus disease incidence per 100,000 person-years, 2002–2006, in 5 different regions of Colorado: the western Colorado high plateau and mountains, the Larimer–Boulder–Jefferson 3-county portion of the northern Front Range (indicated by bold county outlines), the northern Front Range, the southern Front Range, and the eastern Colorado plains. WNV disease incidences are based on combined case and population data for all census tracts within each region falling within a given category for coverage of area with high entomological risk.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 7.
Figure 7.

Spatial distribution of areas with high projected entomological risk of exposure to Cx. tarsalis in the northern Front Range and western Colorado in relation to census tract-based West Nile virus disease incidence per 100,000 person-years, 2002–2006. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

Figure 8.
Figure 8.

Relationship between elevation (x) and mean daily temperature during June–August 2006 (y) along the Poudre River corridor in Larimer County (y = 28.64793 − 0.00538x; ANOVA; F1,6 = 22.76, r2 = 0.791, P = 0.003).

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 78, 4; 10.4269/ajtmh.2008.78.654

*

Address correspondence to Lars Eisen, Department of Microbiology, Immunology, and Pathology, 1690 Campus Delivery, Colorado State University, Fort Collins, CO 80523. E-mail: lars.eisen@colostate.edu

Authors’ addresses: Anna M. Winters, Bethany G. Bolling, Barry J. Beaty, Carol D. Blair, Andrew M. Meyer, Chester G. Moore, and Lars Eisen, Department of Microbiology, Immunology, and Pathology, Colorado State University, Fort Collins, CO 80523, Tel: +1 (970) 491-8386, Fax: +1 (970) 491-8707, E-mail: lars.eisen@colostate.edu. Rebecca J. Eisen, Division of Vector-Borne Infectious Diseases, National Center for Vector-Borne, Zoonotic and Enteric Diseases, Centers for Disease Control and Prevention, Fort Collins, CO 80521. W. John Pape, Colorado Department of Public Health and Environment, Denver, CO 80246.

Acknowledgments: The authors thank the U.S. Forest Service and the cities of Fort Collins and Loveland for permission to collect mosquitoes and S.L. Anderson and J. Montgomery for technical assistance. The study was funded, in part, by a grant from the Centers for Disease Control and Prevention (T01/CCT822307) and a contract with the National Institutes of Health (AI-25489).

REFERENCES

  • 1

    Goddard LB, Roth AE, Reisen WK, Scott TW, 2002. Vector competence of California mosquitoes for West Nile virus. Emerg Infect Dis 8 :1385–1391.

    • Search Google Scholar
    • Export Citation
  • 2

    Turell MJ, Dohm DJ, Sardelis MR, O’Guinn ML, Andreadis TG, Blow JA, 2005. An update on the potential of North American mosquitoes (Diptera: Culicidae) to transmit West Nile virus. J Med Entomol 42 :57–62.

    • Search Google Scholar
    • Export Citation
  • 3

    Turell MJ, Sardelis MR, O’Guinn ML, Dohm DJ, 2002. Potential vectors of West Nile virus in North America. Curr Top Microbiol Immunol 267 :241–252.

    • Search Google Scholar
    • Export Citation
  • 4

    Bolling BG, Moore CG, Anderson SL, Blair CD, Beaty BJ, 2007. Entomological studies along the Colorado Front Range during a period of intense West Nile virus activity. J Am Mosq Contr Assoc 23 :37–46.

    • Search Google Scholar
    • Export Citation
  • 5

    Smith GC, Moore CG, Davis T, Savage HM, Thapa AB, Shrestha SL, Karabatsos N, 1993. Arbovirus surveillance in northern Colorado, 1987 and 1991. J Med Entomol 30 :257–261.

    • Search Google Scholar
    • Export Citation
  • 6

    Gujral IB, Zielinski-Gutierrez EC, LeBailly A, Nasci R, 2007. Behavioral risks for West Nile virus disease, northern Colorado, 2003. Emerg Infect Dis 13 :419–425.

    • Search Google Scholar
    • Export Citation
  • 7

    Tempelis CH, Francy DB, Hayes RO, Lofy MF, 1967. Variations in feeding patterns of seven culicine mosquitoes on vertebrate hosts in Weld and Larimer counties, Colorado. Am J Trop Med Hyg 16 :111–119.

    • Search Google Scholar
    • Export Citation
  • 8

    Tsai TF, Smith GC, Happ CM, Kirk LJ, Jakob WL, Bolin RA, Francy DB, Lampert KJ, 1989. Surveillance of St. Louis encephalitis virus vectors in Grand Junction, Colorado, in 1987. J Am Mosq Contr Assoc 5 :161–165.

    • Search Google Scholar
    • Export Citation
  • 9

    Tsai TF, Smith GC, Ndukwu M, Jakob WL, Happ CM, Kirk LJ, Francy DB, Lampert KJ, 1988. Entomologic studies after a St. Louis encephalitis epidemic in Grand Junction, Colorado. Am J Epidemiol 128 :285–297.

    • Search Google Scholar
    • Export Citation
  • 10

    Reisen WK, Lothrop HD, 1995. Population ecology and dispersal of Culex tarsalis (Diptera: Culicidae) in the Coachella Valley of California. J Med Entomol 32 :490–502.

    • Search Google Scholar
    • Export Citation
  • 11

    Wegbreit J, Reisen WK, 2000. Relationships among weather, mosquito abundance, and encephalitis virus activity in California: Kern County 1990–98. J Am Mosq Contr Assoc 16 :22–27.

    • Search Google Scholar
    • Export Citation
  • 12

    Hess AD, Hayes RO, 1970. Relative potentials of domestic animals for zooprophylaxis against mosquito vectors of encephalitis. Am J Trop Med Hyg 19 :327–334.

    • Search Google Scholar
    • Export Citation
  • 13

    Lothrop HD, Reisen WK, 2001. Landscape affects the host-seeking patterns of Culex tarsalis (Diptera: Culicidae) in the Coachella Valley of California. J Med Entomol 38 :325–332.

    • Search Google Scholar
    • Export Citation
  • 14

    Baker M, 1961. The altitudinal distribution of mosquito larvae in the Colorado Front Range. Trans. Am. Entomol. Soc. 87 :231–246.

  • 15

    Harmston FC, Lawson FA, 1967. Mosquitoes of Colorado. Atlanta, GA: U.S. Department of Health, Education and Welfare, Public Health Service.

  • 16

    Carpenter SJ, 1969. Observations on the distribution and ecology of mountain Aedes mosquitoes in California. XII. Other species found in the mountains. Calif. Vector Views 16 :27–34.

    • Search Google Scholar
    • Export Citation
  • 17

    Darsie RF Jr, Ward RA, 2005. Identification and Geographical Distribution of the Mosquitoes of North America, North of Mexico. Gainesville, FL: University Press of Florida.

  • 18

    Bailey SM, Eliason DA, Hoffman BL, 1965. Flight and dispersal of the mosquito Culex tarsalis Coquillett in the Sacramento Valley of California. Hilgardia 37 :73–113.

    • Search Google Scholar
    • Export Citation
  • 19

    Dow RP, Reeves WC, Bellamy RE, 1965. Dispersal of female Culex tarsalis into a larvicided area. Am J Trop Med Hyg 14 :656–670.

  • 20

    Reisen WK, Milby MM, Meyer RP, Pfuntner AR, Spoehel J, Hazelrigg JE, Webb JP Jr, 1991. Mark–release–recapture studies with Culex mosquitoes (Diptera: Culicidae) in southern California. J Med Entomol 28 :357–371.

    • Search Google Scholar
    • Export Citation
  • 21

    Sall J, Creighton L, Lehman A, 2005. JMP Start Statistics. Third edition. Belmont, CA: Brooks/Cole.

  • 22

    Leung LR, Qian Y, Bian X, Washington WM, Han J, Roads JO, 2004. Mid-century ensemble regional climate change scenarios for the western United States. Clim Change 62 :75–113.

    • Search Google Scholar
    • Export Citation
  • 23

    Diuk-Wasser MA, Brown HE, Andreadis TG, Fish D, 2006. Modeling the spatial distribution of mosquito vectors for West Nile virus in Connecticut, USA. Vector Borne Zoonotic Dis 6 :283–295.

    • Search Google Scholar
    • Export Citation
  • 24

    Zou L, Miller SN, Schmidtmann ET, 2006. Mosquito larval habitat mapping using Remote Sensing and GIS: implications of coalbed methane development and West Nile virus. J Med Entomol 43 :1034–1041.

    • Search Google Scholar
    • Export Citation
  • 25

    Shone SM, Ferrao PN, Lesser CR, Norris DE, Glass GE, 2001. Analysis of mosquito vector species abundances in Maryland using geographic information systems. Ann NY Acad Sci 951 :364–368.

    • Search Google Scholar
    • Export Citation
  • 26

    Shaman J, Stieglitz M, Stark C, Le Blancq S, Cane M, 2002. Using a dynamic hydrology model to predict mosquito abundances in flood and swamp water. Emerg Infect Dis 8 :6–13.

    • Search Google Scholar
    • Export Citation
  • 27

    Rios J, Hacker CS, Hailey CA, Parsons RE, 2006. Demographic and spatial analysis of West Nile virus and St. Louis encephalitis in Houston, Texas. J Am Mosq Contr Assoc 22 :254–263.

    • Search Google Scholar
    • Export Citation
  • 28

    Kronenwetter-Koepel TA, Meece JK, Miller CA, Reed KD, 2005. Surveillance of above- and below-ground mosquito breeding habitats in a rural Midwestern community: baseline data for larvicidal control measures against West Nile virus vectors. Clin Med Res 3 :3–12.

    • Search Google Scholar
    • Export Citation
  • 29

    Gibbs SEJ, Wimberly MC, Madden M, Masour J, Yabsley MJ, Stallknecht DE, 2006. Factors affecting the geographic distribution of West Nile virus in Georgia, USA: 2002–2004. Vector Borne Zoonotic Dis 6 :73–82.

    • Search Google Scholar
    • Export Citation
  • 30

    Theophilides CN, Ahearn SC, Grady S, Merlino M, 2003. Identifying West Nile virus risk areas: the dynamic continuous-area space–time system. Am J Epidemiol 157 :843–854.

    • Search Google Scholar
    • Export Citation
  • 31

    Ward MP, 2006. Spread of equine West Nile virus encephalomyelitis during the 2002 Texas epidemic. Am J Trop Med Hyg 74 :1090–1095.

  • 32

    Ward MP, Ramsay BH, Gallo K, 2005. Rural cases of equine West Nile virus encephalomyelitis and the normalized difference vegetation index. Vector Borne Zoonotic Dis 5 :181–188.

    • Search Google Scholar
    • Export Citation
  • 33

    Brownstein JS, Rosen H, Purdy D, Miller JR, Merlino M, Mostashari F, Fish D, 2002. Spatial analysis of West Nile virus: rapid risk assessment of an introduced vector-borne zoonosis. Vector Borne Zoonotic Dis 2 :157–164.

    • Search Google Scholar
    • Export Citation
  • 34

    Corrigan RLA, Waldner C, Epp T, Wright J, Whitehead SM, Bangura H, Young E, Townsend HGG, 2006. Prediction of human cases of West Nile virus by equine cases, Saskatchewan, Canada, 2003. Prev Vet Med 76 :263–272.

    • Search Google Scholar
    • Export Citation
  • 35

    Yiannakoulias NW, Schopflocher DP, Svenson LW, 2006. Modelling geographic variations in West Nile virus. Can J Publ Health 97 :374–378.

  • 36

    Ruiz MO, Tedesco C, McTighe T, Austin C, Kitron U, 2004. Environmental and social determinants of human risk during a West Nile virus outbreak in the greater Chicago area, 2002. Int J Health Geogr 3 :8–19.

    • Search Google Scholar
    • Export Citation
  • 37

    Ruiz MO, Walker ED, Foster ES, Haramis LD, Kitron U, 2007. Association of West Nile virus illness and urban landscapes in Chicago and Detroit. Int J Health Geogr 6 :10–20.

    • Search Google Scholar
    • Export Citation
  • 38

    Ward MR, Stallknecht DE, Willis J, Conroy MJ, Davidson WR, 2006. Wild bird mortality and West Nile virus surveillance: biases associated with detection, reporting, and carcass persistence. J Wildl Dis 42 :92–106.

    • Search Google Scholar
    • Export Citation
  • 39

    Patnaik JL, Juliusson L, Vogt RL, 2007. Environmental predictors of human West Nile virus infections, Colorado. Emerg Infect Dis 13 :1788–1790.

    • Search Google Scholar
    • Export Citation
  • 40

    Eisen L, Eisen RJ, 2007. Need for improved methods to collect and present spatial epidemiologic data for vectorborne diseases. Emerg Infect Dis 13 :1816–1820.

    • Search Google Scholar
    • Export Citation
  • 41

    Eisen RJ, Eisen L, 2008. Spatial modeling of human risk of exposure to vector-borne pathogens based on epidemiological versus arthropod vector data. J Med Entomol 45 :181–192.

    • Search Google Scholar
    • Export Citation
  • 42

    Lothrop HD, Reisen WK, 1999. A geographical information system to manage mosquito and arbovirus surveillance and control data in the Coachella Valley of California. J Am Mosq Contr Assoc 15 :299–307.

    • Search Google Scholar
    • Export Citation
  • 43

    Takeda T, Whitehouse CA, Brewer M, Gettman AD, Mather TN, 2003. Arbovirus surveillance in Rhode Island: assessing potential ecologic and climatic correlates. J Am Mosq Contr Assoc 19 :179–189.

    • Search Google Scholar
    • Export Citation
  • 44

    Brownstein JS, Holford TR, Fish D, 2004. Enhancing West Nile virus surveillance, United States. Emerg Infect Dis 10 :1129–1133.

  • 45

    Cooke W III, Grala K, Wallis R, 2006. Avian GIS models signal human risk for West Nile virus in Mississippi. Int J Health Geogr 5 :36–56.

    • Search Google Scholar
    • Export Citation
  • 46

    Kutz FW, Wade TG, Pagac BB, 2003. A geospatial study of the potential of two exotic species of mosquitoes to impact the epidemiology of West Nile virus in Maryland. J Am Mosq Contr Assoc 19 :190–198.

    • Search Google Scholar
    • Export Citation
  • 47

    Lukacik G, Anand M, Shusas EJ, Howard JJ, Oliver J, Chen H, Backenson PB, Kauffman EB, Bernard KA, Kramer LD, White DJ, 2006. West Nile virus surveillance in mosquitoes in New York State, 2000–2004. J Am Mosq Contr Assoc 22 :264–271.

    • Search Google Scholar
    • Export Citation
  • 48

    Mongoh MN, Kaitsa ML, Dyer NW, 2006. Environmental and ecological determinants of West Nile virus occurrence in horses in North Dakota, 2002. Epidemiol Infect 135 :57–66.

    • Search Google Scholar
    • Export Citation
  • 49

    Orme-Zavaleta J, Jorgensen J, D’Ambrosio B, Altendorf E, Rossignol PA, 2006. Discovering spatio-temporal models of the spread of West Nile virus. Risk Anal 26 :413–422.

    • Search Google Scholar
    • Export Citation
  • 50

    Eisen RJ, Lane RS, Fritz CL, Eisen L, 2006. Spatial patterns of Lyme disease risk in California based on disease incidence data and modeling of vector-tick exposure. Am J Trop Med Hyg 75 :669–676.

    • Search Google Scholar
    • Export Citation

Author Notes

Reprint requests: Lars Eisen, Department of Microbiology, Immunology, and Pathology, 1690 Campus Delivery, Colorado State University, Fort Collins, CO 80523, Tel: +1 (970) 491-8386, Fax: +1 (970) 491-8707, E-mail: lars.eisen@colostate.edu.
Save