Potential distributions for F. tularensis subspecies and clades across the United States, as predicted by ecological-niche models (), showed that predicted distributions of the two subspecies (types A and B) overlap broadly and cover much of the country. A subtle difference is that type B has a somewhat more northerly distribution than type A. Type B is predicted absent from the southern states of Louisiana, Mississippi, Georgia, South Carolina, and Florida. Conversely, type A is predicted absent from the northern states of Maine, Vermont, and New Hampshire as well as Northern New York, Michigan, and Minnesota. Type B is also predicted to have a more patchy distribution in the Western compared with the Eastern United States.
Potential distribution maps for six tularemia genotypes in the United States. White indicates predicted absence, whereas darker shades correspond to higher model agreement in predictions of potential for presence among the 25 replicate models.
At the clade level, distributions predicted for A1 and A2 contrast sharply. A1 is distributed largely in the Central to Southeastern United States with predicted areas of distribution also in California, Oregon, Northern Utah, and Western Idaho, whereas A2 is concentrated in the Western United States, excepting some regions of Southern Arizona, Western California, Western Oregon, Washington, and Northern Idaho. Finally, A1a and A1b are both predicted to occur across the Central and Southeastern United States, plus California, Oregon, Northern Utah, and Western Idaho.
Histograms summarizing the modeled distributions of F. tularensis subspecies and clades relative to selected environmental and topographical factors () show broad overlap in most variables (annual precipitation, slope, topographic index, diurnal temperature range, minimum temperature, and vapor pressure) for type A and type B. Minor differences in maximum temperature, mean temperature, solar radiation, and wet-day frequency suggest that the more northerly type B distribution is linked to cooler, cloudier regions compared with type A. In contrast, the modeled distributions of A1 and A2 strains differ markedly with respect to several variables (elevation, diurnal temperature range, frost-free days, precipitation, minimum temperature, mean temperature, and vapor pressure). Compared with A1 strains, A2 strains occur at higher elevations (usually > 1,000 m) and in areas with less precipitation, lower temperatures, more ground-frost days, and lower vapor pressure. When comparing A1a and A1b strains, no appreciable difference between modeled distributions was noted with respect to any environmental or topographic variable.
Histograms showing three paired comparisons (first column, type A versus type B; second column, A1 vs. A2; third column, A1a vs. A1b) of the frequency distribution for five environmental variables (rows).
Two-dimensional visualizations of modeled distributions with respect to environmental variables for the different subspecies and clades painted a similar picture. Again, the most dramatic ecological differences were between genotypes A1 and A2 (, middle). Types A and B also showed only minor differences in ecological space, and broad overlap was observed between A1a and A1b.
Scatter plots comparing distributions of F. tularensis genotypes in ecological space based on pairs of environmental variables. Polygons outline the distribution of the general environmental characteristics across the United States.
The MANOVA revealed similar patterns for the ecological-niche comparisons between genotypes. Plotting Fobs and F*/Fobs ratios (), similar ecological niches are expected to fall in the upper left-hand corner of the graph, whereas more different ecological niches fall in the lower right-hand corner. It was clear that subtypes A1a and A1b had the highest values of the ratio combined with lower Fobs values, indicating negligible differentiation between them. Types A and B showed only minor differences. However, the comparison of A1 and A2 showed low values for the ratio, combined with the highest values of Fobs, pointing to highly differentiated ecological niches for these two genotypes.
Figure 5. Scatter plot of the values of the F*/Fobs ratio and the Fobs calculated from the non-parametric MANOVA. Values toward the lower right corner show greater differentiation between the compared genotypes, and values toward the upper left corner show more (more ...)