Geographical and environmental factors driving the increase in the Lyme disease vector Ixodes scapularis

Abstract

The population densities of many organisms have changed dramatically in recent history. Increases in the population density of medically relevant organisms are of particular importance to public health as they are often correlated with the emergence of infectious diseases in human populations. Our aim is to delineate increases in density of a common disease vector in North America, the blacklegged tick, and to identify the environmental factors correlated with these population dynamics. Empirical data that capture the growth of a population are often necessary to identify environmental factors associated with these dynamics. We analyzed temporally- and spatially-structured field collected data in a geographical information systems framework to describe the population growth of blacklegged ticks (Ixodes scapularis) and to identify environmental and climatic factors correlated with these dynamics. The density of the ticks increased throughout the study's temporal and spatial ranges. Tick density increases were positively correlated with mild temperatures, low precipitation, low forest cover, and high urbanization. Importantly, models that accounted for these environmental factors accurately forecast future tick densities across the region. Tick density increased annually along the south-to-north gradient. These trends parallel the increases in human incidences of diseases commonly vectored by I. scapularis. For example, I. scapularis densities are correlated with human Lyme disease incidence, albeit in a non-linear manner that disappears at low tick densities, potentially indicating that a threshold tick density is needed to support epidemiologically-relevant levels of the Lyme disease bacterium. Our results demonstrate a connection between the biogeography of this species and public health.

Keywords: GIS; Ixodes scapularis; blacklegged ticks; density increase; emerging zoonoses; geographic information systems.

Fig. 1

Map of the study area. The locations of tick density estimates are represented as black dots, the Hudson and Mohawk Rivers in a bold line, and the counties of New York State considered in this study are shaded.

Fig. 2

Illustrative representation of the temporal variation and spatial gradient of tick density estimates. (A) Nymphal tick densities were greater in later years and at lower latitudes of the study area, decreasing in northern latitudes. (B) The observed patterns in tick density estimates across space and time are similar to the observed patterns in reported human Lyme disease cases. 4 outliers (very high values; 1 from [A] and 3 from [B]) were removed from figures for graphical clarity but considered for trend lines.

Fig. 3

The nymph and adult density estimate regression models built using data from 2004–2008 accurately predict the tick density estimates in 2009 and 2010. Estimates predicted by the nymphal model explain (A) 80% of the variation in the observed nymph density estimates from 2009 (R²=0.8, n=35, P < 0.0001) and (B) 74% of the variation in the observed tick density estimates from 2010 (R²=0.74, n=26, P < 0.0001). Estimates predicted by the adult model explains (C) 48% of the variation in the observed adult density estimates from 2009 (R² = 0.48, n = 40, P < 0.0001) and (D) 67% of the variation in the observed tick density estimates from 2010 (R²=0.67, n=29, P < 0.0001). Original units prior to log transformation are individuals/hectares. The dotted diagonal line through the origin represents the ideal correspondence (1:1) between observed and predicted density estimates.

Fig. 4

Spatial representation of predictions for nymph (2009 [A] and 2010 [B]) and adult (2009 [C] and 2010 [D]) density estimates. The performance of selected models are shown using the predicted collection values for the region plotted as a surface (original units prior to log transformation are individuals/30-arcsecond pixels) for year 2009 and 2010 and the observed collection values (circles). The observed values represent the average of multiple samples collected in the same year and location. Matching of color inside the circles with the continuous surfaces describes the accuracy of the model predictions.

Fig. 5

The relationship between model-predicted nymphal and adult density estimates (original units prior to log transformation are individuals/hectares) and the observed incidence of Lyme disease per county from 2004 to 2009. (A) Lyme disease incidence is significantly correlated with nymph density estimates greater than the hinge value (dotted line; hinge value=1.13 individuals hectare⁻¹; R²=0.45, n=270, P < 0.0001) but not when estimates are lower than the hinge value (R² = 0.02, n = 40, P > 0.05). (B) Similarly, the correlation between Lyme disease incidence and local adult densities is only significant for values greater that the hinge value (hinge value = 0.71 individuals hectare⁻¹; R² = 0.42, n = 281, P < 0.0001) but not when the estimates are lower than the hinge value (R² = 0.02, n = 29, P > 0.05).