Predicted Northward Expansion of the Geographic Range of the Tick Vector Amblyomma americanum in North America under Future Climate Conditions

Anthropogenic climate change (Cook et al. 2013; IPCC 2018) is likely to drive changes in the geographic ranges of arthropod disease vectors, including those of tick vectors in North America (Ogden et al. 2005; Minigan et al. 2018; Springer et al. 2015). This likelihood is because the survival of tick populations depends on both biotic and abiotic conditions. Temperature plays a critical role in the tick life cycle by determining development rates of eggs and engorged states (Koch 1983) and affecting tick questing activity (Haile and Mount 1987). Subzero air temperatures are not lethal for ticks if they can find refuges in their environment, particularly in the surface layer of the soil (Burks et al. 1996). However, due to its effects on development and activity, temperature determines the length of the tick life cycle. Even where habitats provide refuges from subzero temperatures, a threshold temperature condition occurs below which the tick populations cannot survive, i.e., temperature conditions are too low for the tick to complete its life cycle before it dies, given a particular daily probability that a tick survives (Ogden et al. 2005; Ludwig et al. 2016). Temperature may, therefore, be a limiting factor of the geographic ranges of ticks, and a warming climate may facilitate their establishment in regions previously climatically unsuitable.

Throughout the 20th century, the geographic range of Amblyomma americanum has expanded from the southeastern United States northward to locations in Michigan and New York states that are close to the Canadian border (Springer et al. 2014). This range expansion may have been driven by anthropogenic climate change, which has resulted in a warming trend in the late 20th century in North America (Crowley 2000; Stott et al. 2000; Blunden and Arndt 2019), although there have been no efforts to date to attribute changes in geographic distribution to climate change. This range expansion has had public health impact, at least in terms of increased incidence of spotted fever group rickettsioses (Dahlgren et al. 2016). A. americanum is a recognized public health threat, known for its aggressive host-seeking behavior and vector competence for a wide range of zoonotic pathogens, including Francisella tularensis (the cause of tularemia; Goddard and Varela-Stokes 2009), Ehrlichia chaffeensis (the cause of human monocytic ehrlichiosis; Brouqui 1998), Rickettsia rickettsii (the cause of Rocky Mountain spotted fever; Levin et al. 2017) and Heartland virus (Savage et al. 2016). Recently, it has been suggested that the bite of A. americanum may trigger red meat allergy (Commins et al. 2011).

A number of studies have assessed associations between A. americanum tick population occurrence and density and environmental predictors (Koch and Burg 2006; Schulze et al. 2001; Willis et al. 2012). Studies have also explored the potential effects of climate change on the spatial distribution of the tick (Springer et al. 2015), suggesting possible northward range expansion that may affect northern U.S. states and southern Canada, although A. americanum is not yet considered established in Canada and has not been detected in extensive field surveillance conducted in recent years to track the expansion of I. scapularis ticks (Bouchard et al. 2015). However, in recent years a small number of specimens, likely imported by migratory birds or travelers, were detected in passive tick surveillance, which suggests that if environmental conditions are, or become, suitable in northern U.S. states and Canada, this tick species could become established (Gasmi et al. 2018). By the end of the 21st century, Canada would “very likely” face a mean annual temperature rise in the range of 2–4°C in comparison with current climate (Romero-Lankao et al. 2014) and potentially higher than 5°C under a high greenhouse gas emissions scenario (Ogden and Gachon 2019). If indeed temperature conditions are a major determinant of the northern limit of the range of this tick, then A. americanum may expand its range northward into northern U.S. states and Canada in the future and bring with it the range of diseases and health issues with which it is associated. Consequently, here we evaluate current and future climate suitability for A. americanum in the northern United States and Canada to better prepare public health system for possible emergence of these diseases and health issues.

There are two main modeling approaches to understanding and estimating the relationships between environmental variables, such as climate, and the environmental suitability for arthropods, such as A. americanum. The most commonly used are “pattern matching” approaches, such as ecological niche modeling that infer environmental suitability from observed species-distribution data. This approach has been used to predict current environmental suitability in North America for A. americanum and to project impacts of climate change on the possible geographic distribution of the tick (Bouzek et al. 2013; Springer et al. 2015; Minigan et al. 2018; Kessler et al. 2019; Pascoe et al. 2019). An alternative approach is the use of mechanistic simulation models that identify how environmental factors such as temperature affect population processes in the tick life cycle (Haile and Mount 1987; Mount et al. 1993; Ogden et al. 2005; Ludwig et al. 2016). Both methods have advantages and disadvantages that have been reviewed elsewhere (Mannelli et al. 2016). In previous studies on I. scapularis ticks, simulations of a deterministic simulation model calibrated with temperature-dependent parameters were used to define suitable temperature conditions for establishment of tick populations (Ogden et al. 2005) and then to project future spatial distribution of the species under climate change (Ogden et al. 2006, 2008). A similar dynamic population model was developed by Ludwig et al. (2016), which describes the transformations that A. americanum ticks undergo as they pass through the four stages of their life cycle (egg, larvae, nymph, and adult). Feeding tick stages are hardening, questing, feeding, and engorged ticks; tick abundance in a specific simulated area depends on seasonally variable environmental conditions, including temperature and day length, while accounting for effects of temperature-independent diapause.

In this study, we used this mechanistic model to investigate the relationship between the size of A. americanum tick populations and temperature conditions at 36 locations in southeastern Canada and the eastern United States. From these simulations we defined threshold temperature conditions for tick population survival and evaluated how that the northern limit of the region with suitable climate conditions for the tick in North America may change with climate change.

We ran simulations of the dynamic population model (Ludwig et al. 2016) using climate data from each of 36 meteorological stations located in southeastern Canada and northeastern United States (from latitudes 35 to 60°N, and from longitudes 55 to 95°W) to define a temperature threshold for survival of A. americanum populations. Under current climate, the region where the meteorological stations are located (Figure 1) is characterized by a north–south latitudinal gradient in temperature (being warmer in the south) with temperatures also being higher the closer that the stations are to the Atlantic coastline. The deduced temperature threshold was used to evaluate the tick’s potential range in the United States and Canada under future climatic conditions. Our focus was investigating possible northward range expansion of the tick, but northward contraction of the southern limit of the tick may also be possible. If so then the impact on distribution in the southern United States may be small because the tick is likely established in Mexico at present (Sosa-Gutierrez et al. 2016), but further studies targeted to the southern part of the tick’s range are required to explore this impact.

In this study, using a deterministic population model we sought to identify a lower temperature threshold for persistence of A. americanum populations that may define the northern limit for populations of this tick. We then explored how climate change may affect the northern extent of the geographic range of this tick by northward expansion of the area of climatic suitability in North America. Our results suggest considerable potential for northward range expansion of the tick throughout the 21st century.

Our simulations with modified mortality rates among feeding ticks reproduced results of the study of Ludwig et al. (2016), with respect to suitability of current temperature conditions in southeastern Canada for A. americanum populations, and suggest that the region of current temperature suitability extends into southwestern Quebec and southern Ontario.

The simulation results for meteorological stations in the United States were consistent with A. americanum occurrence based on passive surveillance at eight sites but were not consistent for three sites (Hartford, Connecticut; Franklin, Kentucky; and Franklin, Ohio) that were predicted to be suitable for tick establishment but had no documented evidence of tick presence according to Springer et al. (2014). Evidence of A. americanum populations has been found at locations further north than these three sites (Springer et al. 2014). Therefore, the absence of ticks from the counties in which the sites occur may reflect the influence of environmental factors that were not accounted for by the model, such as suboptimal habitat, relative humidity, or host densities (Koch 1983; Lancaster Jr and McMillan 1955). However, detection of tick populations in Springer et al. (2014) was by passive surveillance, and it may be that there are indeed A. americanum populations present but undetected (Barrett et al. 2015). The predicted current range of climatic suitability in the central and eastern regions of the United States is also consistent with observations. Temperature suitability was also predicted to extend west of the Rocky Mountains. To our knowledge, populations of the tick have not been found here (Springer et al. 2014). Precisely why this absence is the case is not known; other essential environmental factors for A. americanum tick population survival (woodland habitat, relative humidity, host densities) are likely present (Pascoe et al. 2019), so perhaps the mountains and the prairies are a physical barrier to dispersal of this tick. When comparing the predicted range of temperature suitability in North America against surveillance data, Presque Isle county in Michigan is the only county that has been identified as having “established” A. americanum, yet it has temperature conditions predicted to be unsuitable for the tick. However, this county was considered “established” on the basis of passive tick surveillance data (Walker et al. 1998). Similarly, 13 counties in the northern United States identified as positive for A. americanum using the “reported” criteria, were also predicted as having unsuitable temperature conditions. Again, though, for these counties A. americanum populations were detected by passive tick surveillance methods (Means and White, 1997; Keirans and Lacombe, 1998; Walker et al. 1998). Locations identified as positive for tick populations based on passive surveillance may be false positives due to the presence of “adventitious” ticks that can be dispersed long distances out of tick populations by hosts such as migratory birds (Gasmi et al. 2018). In the absence of active field surveillance in these locations it is currently impossible to fully confirm whether tick populations are present in the counties identified as “reported” or “established.” Thus, we considered that the temperature threshold estimated for A. americanum identified here may be able to define the northern limit of the geographic range of the tick, although further field surveillance is needed to confirm this.

The model was sensitive to variations in the mortality rates of feeding and nonfeeding ticks, for which we have no accurate estimates. However, variations in these mortality rates had a very small impact on the estimated lower limit for population survival, which was 3,285 $\text{DD}>0°\mathrm{C}$. This DD threshold is higher than that estimated for I. scapularis (Ogden et al. 2005), which is consistent with the observed spread of I. scapularis tick populations into Canada in advance of A. americanum.

Northward range expansion of A. americanum is expected with climate change on the assumption that cold temperature conditions are the main limiting factor at present. Projected range expansion was not quite as dramatic as that projected for I. scapularis (Ogden et al. 2006). Day length affects host-seeking activity of A. americanum (Haile and Mount 1987), but not that of I. scapularis (Durden et al. 1996), and it remains to be investigated to what extent this aspect may result in differences of range expansion between these species. The predicted range expansion in the United States may have limited public health impact as in general regions of high population density are within the current geographic range of the region of temperature suitability (see https://www.census.gov/history/www/reference/maps/ vs. Figure 4). However, possible range expansion into parts of Canada close to the U.S. border was predicted from Alberta eastward into Newfoundland, that encompass some of the most highly populated parts of this country (https://www150.statcan.gc.ca/n1/pub/91-214-x/2015000/section04-eng.htm#a1). The projected range expansions in the ecological niche model studies of Springer et al. (2015) and Raghavan et al. (2019), were more modest than those here. In the current study, only temperature was considered, although the ecological niche models considered other climatic variables, and indeed these variables may limit range expansion of the tick. However, both models of Springer et al. (2015) and Raghavan et al. (2019) consider the whole range of the tick (rather than focusing on the northern limit) and rather under-estimate environmental suitability in northern states in comparison with the observational data in Springer et al. (2014). The simulation model in our study uses experimental data on temperature effects on tick development rates and activity (and other aspects of the tick life cycle), so the temperature limits identified by the model are “real” inasmuch as those experimental data are representative of A. americanum across its geographic range. If they are representative, the lower temperature limit obtained should be a more accurate than that obtained from ecological niche models for a number of reasons. The ecological niche models of A. americanum use surveillance data that has not been systematically collected and use “presence” data based on passive surveillance, which may be falsely positive due to the presence of adventitious ticks. In addition, the niche models consider a wide range of climatic variables, including temperature, that are assumed to limit the species range, but the assumptions are rarely supported by empirical evidence. For example, Peterson and Raghavan (2017) used the minimum temperature of the coldest month as one bioclimatic variable in an ecological niche model for I. scapularis. Empirical data (Brunner et al. 2012) suggest this variable is unlikely to be limiting for I. scapularis, so its inclusion is likely because it correlates with other environmental variables. However, only greater field surveillance data can determine which method is most accurate at identifying the northern range of the tick.

As anticipated, there was only a slight difference between projections made using the RCP4.5 and RCP8.5 emissions pathways up to the 2060s, but thereafter, projected range expansion was much greater under RCP8.5. All climate models projected future temperatures that would permit northward range expansion, but there was marked variation among the models in the degree of projected warming, particularly in the last decades of the 21st century and when the models used RCP8.5.

At sites with higher temperatures (i.e., $\ge \mathrm{4,000}\text{DD}>0°\mathrm{C}$), simulated tick population abundance declined associated with parameterization of reduced host-finding activity according to observations of Haile and Mount (1987). Therefore, very high temperatures may be detrimental to A. americanum populations. In the ecological niche model studies of Springer et al. (2015) and Raghavan et al. (2019), northward range expansion of the geographic range of A. americanum was accompanied by possible northward contraction of the southern limit of the tick as those regions become hotter than the historical range of climate conditions.

In summary, we have used a simulation model approach to estimate the minimum temperature conditions for the survival of A. americanum populations. Using this information, we have identified the possible current extent of temperature suitability in northern North America. We have also identified the possible northern extent of temperature suitability for this tick according to climate projections for the 21st century. The future extent of the northern limit varied according to the emission scenario and regional climate model. The establishment of A. americanum in northern U.S. states, and in the densely populated parts of southeastern and south-central Canada, is plausible in the coming decades, provided that other aspects of the ecological niche of the tick are present. These results suggest that maintaining systematic passive or active surveillance, or both types, for A. americanum in these regions would be prudent.

This study was partly financed by the Centre pour l’étude et la simulation du climat à l’échelle régionale (ESCER) and the Strategic Research Chair on Hydro-Meteorological Risks under Climate Change, both part of the Université du Québec à Montréal; by the National Sciences and Engineering Research Council of Canada; and by the Public Health Agency of Canada. The authors express special thanks to K. Winger, who provided regional climate model simulations from the CRCM5 model developed at the ESCER center. We acknowledge the World Climate Research Programme's Working Group on Regional Climate, and the Working Group on Coupled Modelling, former coordinating body of CORDEX and responsible panel for CMIP5. We also thank the climate modeling groups (listed in Table 2 of this paper) for producing and making available their model output. We also acknowledge the U.S. Department of Defense ESTCP for its support of the NA-CORDEX data archive.

The authors declare they have no actual or potential competing financial interests.