Background

Dogs in the United States (USA) are hosts to a diverse range of ixodid ticks and can become infected with many of the pathogens transmitted by these vectors. Advances in diagnostic test and recording technologies have led to the creation of a monthly dataset containing county-by-county canine test results from across the USA. The Companion Animal Parasite Council (CAPC) has assembled large datasets of such results from commercial laboratories that provide diagnostic tests for canine exposure to Borrelia burgdorferi, Ehrlichia spp. and Anaplasma spp. [1]. These monthly, county-level CAPC prevalence maps generated interest in the utility of the datasets for assessing seroprevalence norms, forecasting future seroprevalence rates and for identifying trends in canine exposure to this array of tick-borne disease agents. A group of vector ecologists, parasitologists, other biologists and statistical modelers met in Atlanta, GA (June 9–10, 2012) to identify factors that could enhance the accuracy of these predictive models. This report narrates the results of the meeting.

Canine diagnostic test results for exposure to tick-borne pathogens, including B. burgdorferi, Ehrlichia spp. and Anaplasma spp., are of significant interest, not only because canine health is important to pet owners and veterinarians, but also because of the public health importance of many of these infectious disease agents. These tick-borne pathogens are transmitted by two phylogenetically distinct groups of ixodid ticks. Members of the ixodid subfamily Prostriata (Ixodes spp.) transmit agents of granulocytic anaplasmosis (Anaplasma phagocytophilum) and Lyme borreliosis (B. burgdoferi) and are likely to include vectors of a more recently described Ehrlichia muris-like agent in the USA. Members of the subfamily Metastriata (e.g., the genera Amblyomma, Dermacentor and Rhipicephalus) transmit agents of canine and human ehrlichiosis (e.g., E. canis, E. chaffeensis and E. ewingii), canine anaplasmosis (A. platys) and spotted-fever group rickettsiosis (i.e., Rickettsia rickettsii, R. conorii and related Rickettsia spp.).

Large datasets have been assembled from reports of diagnostic test results for canine exposure to B. burgdorferi, Anaplasma spp. and Ehrlichia spp. in the USA. For example, from reports submitted nationwide from 2010–2012, 509,195 (7.2%) of 6,996,197 canine samples were seropositive for B. burgdorferi, 270,168 (4.4%) of 6,192,268 samples were seropositive for Anaplasma, and 111,673 (1.1%) of 6,994,683 samples were seropositive for Ehrlichia[2]. A previous national survey, spanning 2001–2007, reported results from 982,336 diagnostic tests for canine exposure to B. burgdorferi and Ehrlichia spp., and 479,640 tests for canine antibodies to Anaplasma spp., with 5.1%, 0.6% and 4.7% of these samples testing seropositive for B. burgdorferi, Ehrlichia and Anaplasma, respectively [3]. Interestingly, when the canine seroprevalence of B. burgdorferi in the 2001–2007 study was compared to the subsequent prevalence of human Lyme disease, the most commonly reported human vector-borne illness in the USA, canine seroprevalence of B. burgdorferi ≥5.1% was predictive of emergent human Lyme disease in low-incidence counties; a low canine seroprevalence (≤1.0%) was associated with minimal risk for emergent human Lyme disease [4]. A subsequent report, however, underscored the importance of other variables, such as the distribution of competent vector species, for accurate interpretation of these canine diagnostic test data [5].

The overall objective of this CAPC-sponsored workshop was to identify factors that are likely to influence the seroprevalence of canine exposure to tick-borne disease agents in the USA, specifically focusing on the factors and the pathogens for which sufficient data are available, so that these factors could be evaluated for incorporation in mathematical models designed to monitor and to predict spatial and temporal seroprevalence patterns. These preliminary factors provided statisticians some of the critical information needed to begin their model-building procedures.

Approach

Two teams of researchers, from various areas of tick and tick-borne pathogen biology, were assembled and tasked with rational identification of factors thought to be relevant to the canine seroprevalence of pathogens transmitted by prostriate (eight team members) or metastriate (seven team members) ticks (Figure 1). Members of each team were selected based upon diverse areas of expertise in tick biology, tick-borne disease, vector ecology or statistics. Each panel was asked to identify and to rank ten key factors that they considered most likely to affect pathogen seroprevalence, and these preliminary factors were then presented to all of the meeting participants for further discussion. It was understood that the relevance of these factors would be subsequently assessed with mathematical models, and that these models would be adjusted with data that continue to be generated. Thus, the utility of different factors would be continually assessed as the mathematical models are refined over time.

Figure 1
figure 1

Approach to rational identification of quantitative factors proposed to influence the exposure of dogs to vector-borne pathogens. Background information, meeting objectives and guidelines were presented to participants before they were divided into three separate panels, according to vector taxa, for mosquitoes, prostriate ticks and metastriate ticks. Each panel was asked to identify, discuss and to rank candidate factors for evaluation with statistical prevalence models of pathogens transmitted to US dogs by each vector taxon. All of the participants were subsequently reconvened for further discussion and refinement of the results from each panel.

Full size image

The working groups for both ixodid subfamilies began by discussing variables categorized as (1) vector, (2) host, (3) abiotic, (4) habitat or (5) social. Both groups independently identified numerous factors. The majority of factors were thought to be associated with canine exposure to pathogens vectored by either ixodid subfamily; however, several factors specifically associated with the different ixodid subfamilies also emerged. Variables were also discussed for which there is little or inconsistent supporting data, but these factors could become useful if the data became available. However, in accordance with the workshop objectives, factors for which sufficient data are currently available were chosen for ranking by consensus of each working group.

The variables independently identified by each panel were categorized into the five groups previously indicated (i.e., vector, host, abiotic, habitat and social). Factors regarding exposure to infectious agents transmitted by prostriate ticks were heavily influenced by the preponderance of research on the phenology of Ixodes scapularis and I. pacificus, which are considered the primary vectors of B. burgdorferi and A. phagocytophilum in North America (Table 1). For metastriate-borne pathogens, host biology and human behavior were second only to vector distribution with regard to factors considered likely to influence seroprevalence (Table 2). Brief explanations and comments regarding these factors are described below.

Table 1 Factors initially considered as potential contributors to canine prevalence of disease agents transmitted by Ixodes scapularis and I. pacificus
Full size table
Table 2 Factors discussed as potential contributors to seroprevalence of metastriate tick-borne pathogens among dogs in the USA
Full size table

Vector factors

Distribution

The geographic distribution of prostriate ticks was focused on the Ixodes spp. thought to most commonly feed on dogs (and people) in the USA: I. scapularis and I. pacificus. Metastriate ticks considered as pathogen vectors (e.g., of Ehrlichia spp. and A. platys) included, in alphabetical order, A. americanum, A. maculatum, D. andersoni, D. variabilis and R. sanguineus. The general distributions of these ticks are relatively well documented in the literature and via voucher specimens in the USA. However, the spatial resolutions of these data vary in different regions, and defining the minimum useful scale can be complicated by discontinuous geographic distributions of tick populations in a given area.

Abundance

Defining permanent values of tick abundance levels is problematic, because tick population levels within a given area are temporally and spatially variable and can change rapidly. Tick abundance depends on host abundance and availability, relative humidity, precipitation and temperature, and can reflect conditions from previous years when immature tick stages or prior generations were active.

Activity

Activity is indicative of questing behavior, host-seeking behavior, host contact and the feeding preferences of different developmental stages. The presence of ticks in an area is not alone indicative of activity. For example, tick activity will depend on temperature, precipitation, relative humidity and photoperiod.

Host factors

Deer

The deer population is a major driver of abundance for certain ticks, such as I. scapularis, I. pacificus and A. americanum. Deer are also a reservoir of E. chaffeensis and could be involved in the maintenance of E. ewingii.

Small mammals

Rodents are an important component of the ecologies of several tick species and some tick-borne infectious agents. Immature stages of several tick species acquire blood meals from small vertebrate hosts. Several tick-borne infectious agents, such as B. burgdorferi, A. phagocytophilum and R. rickettsii are adapted to rodent reservoir hosts.

Lizards

Small vertebrates such as lizards, which are permissive hosts for immature tick stages but are not definitively documented reservoirs of the pathogens under consideration, could dampen transmission of disease agents that are adapted to rodent reservoirs. Conversely, removal of lizards reportedly reduced nymphal tick numbers from an environment but did not affect the percentage of B. burgdorferi-positive ticks, suggesting that increased numbers of lizard hosts might actually increase the risk of pathogen transmission by serving to increase the overall number of ticks in a given area [6].

Migratory bird patterns

Migratory birds can introduce some tick species to new areas [7]. However, ticks that feed on dogs and that are dispersed by birds in the USA may be incapable of maintaining an active population cycle in the absence of larger vertebrate hosts (e.g., white-tailed deer).

Abiotic factors

Different tick species and their natural hosts can be adapted to various environments that are influenced by abiotic factors such as precipitation, temperature, relative humidity and soil composition.

Habitat factors

Factors that influence the life cycles of ticks and their vertebrate hosts include vegetation, urbanization, land use in non-urban settings and detritus layers.

Social factors

Human behavior and population characteristics influence the exposure of dogs to ticks. These include access to preventive care, recreation, socioeconomic status, income, pathogen reservoir control, vector-amplification host control and news media coverage.

Unquantified variables

A number of variables were discussed for which comprehensive, nationwide data did not seem currently available. These variables included vector infection rates, detailed reservoir infection rates, vector abundances, vector efficiency indices, vector survival, vectorial capacities, temperature-dependent development rates of vectors (natural temperature regimes), total number of dogs (by county or zip code) and tick control product sales in each geographic region. Local data may be available for some of these variables in certain areas, but national datasets were not available at the time of this meeting.

Mathematical modeling

Each expert panel was asked to prioritize 10 factors most expected to drive a reliable mathematical predictive model. These lists, summarized in Tables 3 and 4, shared several common abiotic and habitat factors. Several other factors were specific to seroprevalence of the pathogens transmitted by Ixodes spp., R. sanguineus or the other metastriate ticks that were considered. For example, while deer populations and vegetation were considered important factors that affect the majority of these tick populations, social factors were given the highest priority for predicting the seroprevalence of agents transmitted by the brown dog tick, R. sanguineus (Table 4).

Table 3 Ranked factors identified for canine seroprevalence models of infections transmitted by Ixodes spp. in the USA
Full size table
Table 4 Ranked factors for preliminary models of metastriate tick-borne pathogen prevalence among dogs in the USA
Full size table

The prevalence data at the foundation of this predictive model is largely based on serodiagnostic tests. Although seropositivity is reflective of past exposure, it does not demonstrate recent or active infections. Repeatedly seropositive samples from the same dogs at different times are also to be occasionally expected, because some dogs may have tested seropositive in previous tests and because some tests are conducted to monitor host responses to treatment. Travel histories and certainties of the individual test results are currently unavailable for the dogs reported in this dataset.

An analogous project for mathematical modeling of the prevalence of canine heartworm was simultaneously undertaken by CAPC [8, 9], and each prioritized factor identified by the expert panel had significant predictive power with ≥95% confidence. Overall, the model explained 60%-70% of variability in the CAPC county-by-county dataset from 2011–2013. Similarly, preliminary analysis of canine seroprevalence of Anaplasma spp. indicated that temperature, precipitation, relative humidity, population density, median household income, forestation coverage, elevation and deer/vehicle strike rates were significant with ≥95% confidence, and that the total proportion of variability explained in the 2011–2013 data is around 60-70% [10]. Thus, the prevalence of heartworm and seroprevelance of Anaplasma among dogs appear amenable to quantification that could facilitate monitoring for outbreaks, remediation of vector abundance or for forecasting future seroprevalence levels.

Attempts to fit the seroprevalence of B. burgdorferi and of Ehrlichia spp. among dogs are also underway, with mixed results. The spatial seroprevalence of B. burgdorferi among dogs has been similar to and appears to be as quantifiable as that of Anaplasma spp. Conversely, the canine seroprevalence of Ehrlichia spp. appears to be highly variable, with some neighboring areas reporting antipodal seroprevalence rates that could be reflective of vector ecology or social factors. Future work will address these issues.

Conclusions

This meeting brought together a range of junior and senior scientists engaged in various aspects of research in the biology of ticks and tick-borne infections. The specific objectives were to identify and to prioritize quantifiable factors expected to contribute to canine exposure to organisms transmitted by the two major subfamilies of ixodid ticks. The two panels ranked 12 and 17 factors associated with prostriate and metastriate ixodid ticks, respectively. Eight of these factors were independently prioritized by both panels; four of 12 factors were unique to prostriate-vectored agents, two of 11 factors were unique to metastriate-vectored agents transmitted by ticks other than R. sanguineus, and four of six factors were unique to agents vectored by R. sanguineus. The next phase of this project will move from rational identification of perceived factors to statistical assessment of factors for predictive power. Forecasting issues will also be explored.