The locations of the 138 sites that were sampled in Wisconsin, Illinois, and Michigan are shown in Figure 1. Among the four categories, 56 sites were classified as negative, 24 were ranked as 1, 32 as 2, and 26 as 3. Most negative sites were in northeastern Wisconsin. In the southeastern part of Wisconsin, sites were negative except those situated in the Kettle Moraine State Forests (Sheboygan, Fond du Lac, Jefferson, Walworth, and Waukesha counties), which are located on the terminal glacial moraines. Negative sites in Illinois were at Blackhawk Nature Preserve (Rock Island County), located in a suburban area, and White Pines State Park (Ogle County), which has large stands of secondary growth pine forest. In Wisconsin, positive high-density sites were found in the southwestern driftless area and in the central sandy uplands, as well as in the well-recognized northwest part of the state (and across the state line into Minnesota).
Geographic distribution of study sites ranked by abundance of Ixodes scapularis in Wisconsin, northern Illinois, and Menominee County in Michigan.
In Michigan, where only a small area of the Upper Peninsula was sampled, all sites had very dense tick populations, except for a site that was classified as excessively drained (>99% sand). The sites classified in the other two abundance categories (1 and 2) did not appear to cluster in any areas. In Illinois, the two parks that have been infested for at least a decade, Castle Rock State Park (Ogle County) and Loud Thunder Forest Preserve (Rock Island County), were classified as having dense tick populations, with lower populations in some sites along the Illinois River.
Particle size analysis, which is a function of the proportions of sand, silt, and clay, was performed at 82 sites (Figure 2). The positive sites were clustered in the sand/loamy sand texture classes. Individual percentages of sand, silt, and clay per sample were not correlated with tick abundance; however, texture class, which is a combination of these three percentages, correlated significantly (r=0.42, p<0.05) with greater tick densities found in soils with a greater proportion of sand. The soil texture class of samples determined from the soil analysis correlated significantly (r=0.46, p<0.001) with the soil texture class of each site as obtained from the STATSGO database.
Soil particle size analysis of samples from positive and negative sites. Soil texture is expressed as the sum of percent sand, silt, and clay.
The univariate analysis detected significant associations (p<0.25) between tick presence and land cover, soil order, bedrock geology, quaternary geology, soil texture, forest type, spring, summer, fall and winter precipitation, snowfall, and elevation (Figure 3). The results of the discriminant analysis are listed in Table 1 . When negative and positive sites were contrasted, the variables forest type, soil order, land cover, soil texture and bedrock were significant. Tick presence was positively associated with deciduous (Figure 3a), dry/mesic and dry forests (Figure 3b), fertile soils such as alfisols (Figures 3c, Figure 4), sand and loamy/sand soil texture (Figures 2, 3d), and sedimentary bedrock (Figure 3e). There was a negative association with grasslands and conifer forests (Figure 3a), wet and wet/mesic forests (Figure 3b), acidic soils such as spodosols (Figure 3c), clay soil texture (Figure 3d), and Precambrian bedrock (Figure 3e). Elevation was not an important discriminator in the model, nor was Quaternary geology (Figure 3f) important even though sites located on the plateaus and loess-covered areas were all positive. However, the distribution of the sites among the categories of Quaternary deposits was skewed because a large proportion of the state parks were located on terminal glacial moraines. The discriminate model was able to correctly classify 85.7% of the sites. The canonical correlation coefficient was 0.69, and the eigenvalue was close to 1 (0.91), indicative of a strong discriminant function. When the single stage category was included with the negative group, only two variables, forest type and soil order, were significant. Most of the sites (78.6%) were still correctly classified; however, the eigenvalue decreased to 0.43. These same variables were significant when all the groups were considered separately; but the model only correctly classified 51.8% of the sites. Even though only 4/33 in the negative group were misclassified, there was very poor discrimination among the tick positive groups. No significant variables resulted from the discriminant analysis performed using the satellite data. Since all sites were located in forested areas, TM imagery may not have been able to discriminate well among suitable and unsuitable forested habitats. The precipitation variable was also not a significant discriminator between positive and negative sites in the model.
The results of the logistic regression analysis were in agreement with the discriminant analysis model in the positive versus negative group as seen in Table 2 . The same variables were significant (p<0.05), and the model correctly classified 83.9% of the sites. The predictive risk map generated from the logistic regression model is shown in Figure 5. The higher probabilities indicate increased suitability of habitat for I. scapularis. In Wisconsin, the areas of moderate suitability (26%-40%) are located in the western half of the state. Patchy areas of higher probability (60%-100%) are found in the central and northern portion (Juneau, Adams, Waushara, and Marquette counties.) and along the border with Minnesota (Vernon and Crawford counties). In Illinois, the positive sites that were sampled corresponded to areas of increased suitability (60%-100%). Castle Rock State Park, where the highest tick densities are found, had a 90%-100% probability of suitable habitat. The areas bordering the Illinois River appear to be adequate habitat for I. scapularis, especially on the western side. Shawnee National Forest in the extreme southern portion of the state also appears to have a high probability (60%-80%), even though I. scapularis populations have not been detected.
Emerging Infectious Diseases. 2002;8(3) © 2002 Centers for Disease Control and Prevention (CDC)
Cite this: Predicting the Risk of Lyme Disease: Habitat Suitability for Ixodes scapularis in the North Central United States - Medscape - Mar 01, 2002.