Rocky Mountain spotted fever (RMSF), part of a family of diseases known as spotted fever group rickettsioses (SFGR), is recognized as one of the most prevalent and severe tick-borne rickettsial illnesses in the United States (U.S.). RMSF continues to be an emerging public health threat in the south-western United States since its recognition in the region in 2003 (Biggs et al., 2016; Demma et al., 2005). The disease is caused by the highly pathogenic intracellular bacterium Rickettsia rickettsii, with early nonspecific clinical manifestations including fever, headache and rash, which can progress to severe sepsis, widespread vasculitis and death. Doxycycline is the recommended antibiotic therapy and is most effective when administered early in the course of illness (Biggs et al., 2016; Demma et al., 2005; Openshaw et al., 2010).

The emergence of RMSF in Arizona in 2003 (McQuiston et al., 2014) led to the discovery of Rhipicephalus sanguineus sensu lato, the brown dog tick, as a newly identified vector for RMSF in the United States. (Demma et al., 2005). This finding was supported by a series of investigations in eastern Arizona that revealed a high level of R. rickettsii among Rh. sanguineus s.l. ticks (Demma et al., 2005; Eremeeva et al., 2006; Nicholson, Gordon, & Demma, 2006; Nicholson, Paddock, et al., 2006). The ecology of Rh. sanguineus s.l. is distinct from other tick vectors, and the epidemiology of RMSF in Arizona is unique compared to the rest of the United States, including infection in peridomestic settings, younger case age and higher case fatality rates (Demma et al., 2006, 2005; Openshaw et al., 2010). Since the discovery of RMSF in Arizona, over 385 human cases and 23 fatalities have been identified (ADHS et al., 2017). These cases have predominantly been identified on American Indian tribal lands, where large populations of free-roaming dogs support Rh. sanguineus s.l. populations (Demma et al., 2006; Diniz et al., 2009; Folkema, Holman, McQuiston, & Cheek, 2012; Holman, McQuiston, Haberling, & Cheek, 2009; Nicholson, Gordon, et al., 2006; Nicholson, Paddock, et al., 2006). In addition, binational cases have been identified among Arizona residents that become infected with RMSF while visiting Mexico, where the pathogen and vector are also present (Drexler et al., 2017).

RMSF is re-emerging in Mexico, with multiple recent outbreaks in the states of Sonora, Sinaloa, and Baja California and an increasing number of cases in the border cities of Nogales and Agua Prieta along the Arizona-Sonora border region (Alvarez-Hernández, 2010; Alvarez-Hernández & Contreras Soto, 2013; Alvarez-Hernández et al., 2017; Alvarez-Hernández, Murillo-Benitez, Candia-Plata Mdel, & Moro, 2015). The epidemiologic and clinical features of cases in Mexico parallel what is seen on Arizona American Indian tribal lands, with primarily peridomestic transmission and higher incidence among children. Increased risk for RMSF exposure occurs predominantly in lower socio-economic populations; this is further influenced by increased tick harbourage areas, and an abundance of free-roaming dogs that disperse infected Rh. sanguineus s.l. around human residences (Alvarez-Hernández, 2010; Alvarez-Hernández et al., 2017; Eremeeva et al., 2011; Openshaw et al., 2010).

Dogs are thought to play a major role in the emergence and continual circulation of Rickettsia spp. on Arizona American Indian tribal lands and in Northern Mexico. Dogs are the preferred host for the Rh. sanguineus s.l. tick in all tick life stages and are at high risk for exposure to RMSF in hyperendemic areas (Demma et al., 2006, 2005; Diniz et al., 2009; Elchos & Goddard, 2003; Gasser, Birkenheuer, & Breitschwerdt, 2001; Nicholson, Gordon, et al., 2006; Nicholson, Paddock, et al., 2006). Dogs are clinically susceptible to RMSF and develop similar signs as those exhibited by humans, including fever, myalgia, lymphadenopathy, oedema of the face or extremities and petechial rash (Gasser et al., 2001; Greene & Breitschwerdt, 2011; Nicholson, Allen, McQuiston, Breitschwerdt, & Little, 2010; Paddock et al., 2002). However, infection can also be asymptomatic in dogs. Asymptomatic infection can result in transmission of the bacterium to uninfected ticks that attach (Elchos & Goddard, 2003; Nicholson et al., 2010; Nicholson, Gordon, et al., 2006; Nicholson, Paddock, et al., 2006). While not well-documented, dogs with RMSF infection via a single tick bite mount an antibody response that can persist for up to a year; this particularly applies to dogs that have been exposed in areas of high tick activity (Nicholson et al., 2010). Knowledge of seropositivity rates in dogs and monitoring their trends over time can provide clues for human disease risk, as dogs are likely to be infected before human cases occur. Impacted communities have previously reported an increase in recent illnesses in or deaths of dogs around the home within the weeks prior to human illness (Diniz et al., 2009; Elchos & Goddard, 2003; Nicholson, Gordon, et al., 2006; Nicholson, Paddock, et al., 2006; Paddock et al., 2002). In this way, dogs can serve as an early warning system for RMSF emergence in new areas and help target prevention strategies.

Previous canine serologic surveys conducted in 2003–2004 revealed high levels of antibodies to Rickettsia in dogs from two neighbouring American Indian communities experiencing an outbreak of RMSF (Demma et al., 2006, 2005 ). Additional studies in 2005–2006 outside documented outbreak areas in Arizona evaluated dogs for exposure to SFGR and found an overall 5.7% (range: 0%−17.5%) seropositivity rate (McQuiston et al., 2011). At present, the burden of SFGR along the Arizona-Sonora border region is not well-studied (Diniz et al., 2009; Drexler et al., 2017; McQuiston et al., 2011; Openshaw et al., 2010). Infected ticks and rickettsemic dogs could be transported across the border and possibly be a source of RMSF introduction and expansion to nonendemic areas (Alvarez-Hernández et al., 2017; Demma et al., 2005; Eremeeva et al., 2011; Folkema et al., 2012; Fritz, 2009). The objective of this investigation was to assess seropositivity rates to SFGR in dogs along the Arizona-Sonora border region to determine human risk and improve early detection of human cases.

Serum samples were collected from 217 dogs from Cochise (n = 41), Santa Cruz (n = 106) and Yuma (n = 70) counties (Table 1). The majority of dogs across all sites were adults (>1 year of age) (n = 185; 85%), male (n = 111; 51%) and reported to be stray rather than owned or relinquished (n = 123; 57%). Neuter status varied the most across sampling sites, with 41% spayed or neutered across all sites, and a range of 29%−44% at the individual sites (Table 1). Overall, 11 dogs (5.1%) had positive IgG titres reactive with R. rickettsii; with titres ranging from 64 to 1,024 and a geometric mean titre of 120 (CI=73–197). The reciprocal titres for all dogs sampled ranged from ≤32 to 1,024. Seropositivity across the three counties ranged from 2.9% to 12.2% (Table 2). A higher proportion of seropositive dogs were identified in Cochise County (Table 2), but this finding was not significantly different from Santa Cruz and Yuma Counties. Figure 1 illustrates the locations of seropositive dogs by sampling region. Large dogs were significantly more likely to be seropositive (11/151; 7.3%) than small dogs, as no small dogs had positive serologic results. There were no statistically significant associations between seropositive status and dog’s age, sex, neuter status or ownership status (Table 3).

Owned dogs (n = 94) comprised 43% of the total dogs sampled; 5 (5.3%) were seropositive for R. rickettsii antibodies. Among all owned dogs, 33 had been taken to Mexico as reported by the owner. Areas of frequent travel included Nogales and Algodones, Mexico. Dogs travelled with owners most commonly on a monthly basis (46%). Of the five owned dogs that were seropositive, only one was reported to have recent travel to Mexico, approximately 1–3 months prior to sampling. To assess whether owned dogs living in the border region had any increased risk for tick exposure, owners were asked whether their dogs mostly spent time inside the house and if their dog was contained by a leash or fence when outside. The majority of owned dogs lived inside the house (81%) and was contained by a leash or fence when allowed outside the house (84%). All of the owned dogs that were seropositive were reported to be kept mostly indoors.

Among all dogs in the project, 17 (8%) had tick infestations at the time of sampling. None of these dogs had serologic evidence of exposure to SFGR. Stray dogs were 5.7 times) more likely to have ticks attached than owned dogs (95% CI 1.34–24.45; p < 0.001). Other factors possibly associated with the presence of ticks, including age, sex, size and neuter status, were not statistically significant between dogs with and without ticks (Table 4). The average tick count among dogs with ticks was 3.3 ticks per dog (range 1–15). A total of 57 ticks were collected during the investigation and all ticks were identified as Rh. sanguineus s.l. Four ticks (Figure 1) (7%) had DNA evidence of genera-wide Rickettsia species; however, none of the ticks yielded positive PCR results in the R. rickettsii-specific assay. Among stray dogs, six dogs sampled at one of the study sites were identified to be free-roaming dogs from American Indian tribal lands not known to be impacted by RMSF; all tested seronegative for SFGR antibodies.

This investigation aimed to address the prevalence and distribution of SFGR antibodies in canine populations in the ArizonaSonora border region to better understand the risk to human and animal health. The overall 5.1% seropositivity of dogs identified in this investigation demonstrates seroprevalence levels much lower than reports from highly endemic areas, which ranged from 12.5% to 100% in dogs from impacted American Indian tribes (Demma et al., 2006, 2005; McQuiston et al., 2014, 2011). The level of canine seroprevalence to SFGR found in this study is, however, similar to what was observed in nonendemic areas of Arizona (5.7%) (McQuiston et al., 2011), as well as on American Indian tribal lands in 1996 (5%) before RMSF was identified in humans (Demma et al., 2006, 2005; Folkema et al., 2012; Nicholson, Gordon, et al., 2006; Nicholson, Paddock, et al., 2006; Openshaw et al., 2010). Continued surveillance among dogs and humans for illness compatible with RMSF remains critical to identify any evidence of disease introduction or spread. Serosurveys may specifically provide an effective method of active surveillance to monitor SFGR activity in a defined area.

Among the dogs that had evidence of exposure to SFGR, the only factor associated with seropositivity was size (Table 3), with a higher proportion of large dogs (7.3%) having antibodies compared with small dogs (0%). This finding is potentially an artefact of the overall low level of seropositivity, apparent low levels of tick exposure and small proportion of small dogs sampled in comparison with large dogs. Size might also represent the likelihood of the dogs being regularly housed indoors versus outdoors, and by association risk for tick exposure and seroprevalence. There was no difference in risk for SFGR exposure by ownership status, age, sex, neuter status, presence of ticks or sampling location (Table 3). In a surprising manner, dogs with reported history of travel to areas with evidence of RMSF, such as American Indian tribal lands and Mexico, did not show an increased risk for disease or presence of ticks compared with dogs coming from other areas. The only factor associated with the presence of ticks on the dog was ownership status (Table 4), with a higher proportion of stray dogs having ticks present (12%) compared with owned dogs (2%). None of the dogs with ticks observed at the time of sampling were seropositive. Ownership status likely contributes to dogs being regularly groomed and having tick prevention products applied, decreasing the risk for ticks. Moreover, pet ownership practices may differ in RMSF endemic areas, as dogs sampled during this project were more commonly kept leashed or in fenced yards which can be a protective measure for tick exposure. While ticks were collected from stray dogs, it is also likely that dogs sampled at animal shelters had lower tick burdens than expected due to veterinary care (i.e. tick removal, treatments or baths) received upon intake.

Although this serosurvey indicates limited exposure to SFGR among dog populations in southern Arizona, it does not quantify the risk for disease from ticks transported by dogs directly from Mexico or tribal lands. For example, a fatal case of RMSF reported in California occurred in a 53-year-old female who was exposed to infected ticks that were on a dog brought over from an endemic area in Mexico (Drexler et al., 2017). While trans-national exposure may be rare, this case exemplifies the need for increased education about the risk of RMSF on both sides of the border and the importance of treating dogs with tick preventive products before travel.

The authors recognize several limitations to this investigation. Cross-reactions are possible between antibodies for R. rickettsii and other SFGR family members (Greene & Breitschwerdt, 2011). Serologic evaluation of dogs for other SFGR was not conducted as part of this study, and it is possible that the antibody titres detected were caused by exposure to other closely related SFGR (e.g. R. rhipi- cephali, R. massiliae, or R. parkeri) (Eremeeva et al., 2006; Nicholson et al., 2010; Herrick, et al, 2016; Allerdice, et al, 2017) that have been identified in the state. Also, the convenience sampling design through rabies clinics and animal care facilities might not be representative of all dogs in the border region. Owned dogs which presented at rabies vaccine clinics likely have a different risk for exposure than unowned dogs, or owned dogs not regularly vaccinated for rabies; however, we did not observe any difference in risk for SFGR infection between owned and stray dogs. In addition, because ill dogs were excluded from the study, it is possible that either tick-infested dogs or those with possible infection were not included in the sampling. At last, antibodies represent a cumulative response to single exposures and multiple exposures to both heterologous and homologous rickettsial antigens. IgG antibodies against Rickettsia spp. can persist for lengthy time periods after initial infection, and a positive result does not indicate when infection occurred; conversely, antibodies levels could wane after initial exposure and some dogs could have negative results that were previously infected. In hyperendemic areas, repeated exposures result in high titre values and increased prevalence of antibodies, while other areas have sporadic elevated titres (Demma et al., 2006; McQuiston et al., 2011; Nicholson et al., 2010).

This study provides additional information about the potential geographic distribution of SFGR in the southern Arizona border region. While canine seroprevalence in this investigation was low, these findings indicate that dogs in the region have been exposed to SFGR, as seen in all three counties sampled. Although these data indicate a relatively low risk of RMSF infection to humans, other tick vectors infected with Rickettsia spp. could pose a risk to human health (Nicholson et al., 2010). Because of the growing importance of R. sanguineus s. l. as a vector of Rickettsia spp., continued surveillance for tick populations and symptomatic cases of disease among canine or human populations is needed.

Previous studies (Demma et al, 2006; Diniz et al., 2009; McQuiston et al., 2011) describe associations between canine se- roprevalence and human cases, and provide evidence that dogs can be effective sentinels for RMSF. Antibody seroprevalence investigations of SFGR among dogs in high-risk regions for RMSF introduction can provide valuable evidence about the potential human risk and geographic distribution where risk levels are unknown. Given the nonspecific clinical signs and lack of rapid diagnostic tests for RMSF for canine and human populations, there is a need for empiric antibiotic treatment for dogs or humans with clinically compatible illness and a high degree of suspicion in known endemic areas.

The current study helped to better define the risk of SFGR in the border region, and additional investigations in future years may be beneficial to identify early emergence or spread of SFGR into new geographic boundaries. This project provided an opportunity to educate animal control staff and pet owners and highlighted the importance of education, particularly for owners that travel with their dogs to hyperendemic areas. Continued outreach about the risk of disease and prevention methods is critical to detect and prevent further spread of SFGR in Arizona. Perhaps most importantly, testing and education will lead to greater knowledge and awareness of the disease throughout the border region, tools for prevention and guidance for more effective disease control strategies.