Examining Prevalence and Diversity of Tick-Borne Pathogens in Questing Ixodes pacificus Ticks in California

ABSTRACT Tick-borne diseases in California include Lyme disease (caused by Borrelia burgdorferi), infections with Borrelia miyamotoi, and human granulocytic anaplasmosis (caused by Anaplasma phagocytophilum). We surveyed multiple sites and habitats (woodland, grassland, and coastal chaparral) in California to describe spatial patterns of tick-borne pathogen prevalence in western black-legged ticks (Ixodes pacificus). We found that several species of Borrelia—B. burgdorferi, Borrelia americana, and Borrelia bissettiae—were observed in habitats, such as coastal chaparral, that do not harbor obvious reservoir host candidates. Describing tick-borne pathogen prevalence is strongly influenced by the scale of surveillance: aggregating data from individual sites to match jurisdictional boundaries (e.g., county or state) can lower the reported infection prevalence. Considering multiple pathogen species in the same habitat allows a more cohesive interpretation of local pathogen occurrence. IMPORTANCE Understanding the local host ecology and prevalence of zoonotic diseases is vital for public health. Using tick-borne diseases in California, we show that there is often a bias to our understanding and that studies tend to focus on particular habitats, e.g., Lyme disease in oak woodlands. Other habitats may harbor a surprising diversity of tick-borne pathogens but have been neglected, e.g., coastal chaparral. Explaining pathogen prevalence requires descriptions of data on a local scale; otherwise, aggregating the data can misrepresent the local dynamics of tick-borne diseases.

Here, we examined the prevalence patterns of B. burgdorferi, B. americana, B. miyamotoi, B. bissettiae, and A. phagocytophilum in questing I. pacificus ticks at sites in coastal counties of central and northern California. We also explored the impacts of aggregating data from site to regional levels.

RESULTS
B. burgdorferi sensu lato prevalence in adult ticks. Collection sites and infection prevalence for B. burgdorferi sensu lato and Borrelia miyamotoi are shown in Fig. 2. Aggregated across all sites, real-time PCR prevalence of B. burgdorferi sensu lato was 2.9% (95% confidence interval [CI] = 2.3 to 3.7%) in adult ticks ( Table 2). A total of 36 B. burgdorferi sensu lato samples were successfully sequenced from adult ticks, of which 32 (89%) were B. burgdorferi sensu stricto (Table 3). Sequencing also identified the presence of B. americana (n = 3) and B. bissettiae (n = 1) ( Table 3; Fig. 3); further description of these results is below. Adult tick populations from Marin, Monterey, Napa, Sonoma, and Santa Cruz counties all harbored B. burgdorferi sensu stricto (sample sizes were .73 for each county). We did not observe B. burgdorferi sensu lato in adult ticks collected in Mendocino or Santa Clara counties, though samples from these two counties were small (n , 24 for both counties), and ticks were predominantly collected from coastal grassland or chaparral habitats in Mendocino County.
Several individual sites exhibited B. burgdorferi sensu lato prevalence greater than 3.7% in adult ticks, i.e., higher than the confidence intervals generated when all the data were aggregated (2.3 to 3.7%). Some of these sites included different species of B. burgdorferi sensu lato, such as B. burgdorferi sensu stricto, B. americana, and B. bissettiae (e.g., Marin Headlands and Tennessee Valley), and some were in habitats not traditionally associated with B. burgdorferi sensu lato (e.g., chaparral or redwood forest). We did not observe B. burgdorferi sensu lato in adult I. pacificus ticks at 11/27 of the sites at which we tested .30 ticks (11 sites representing 768 ticks).
At individual sites where B. miyamotoi was present (and where samples sizes were .30; n = 9), the prevalence of B. miyamotoi often exceeded the prevalence defined by the confidence intervals generated by the aggregated data (0.8 to 1.8%), ranging from 2.2 to 3.9% at 6 sites (Tables 2 and 3). Like B. burgdorferi sensu lato, B. miyamotoi was not observed in Mendocino or Santa Clara counties, where samples were small.  B. burgdorferi sensu lato prevalence in nymphal ticks. In total, B. burgdorferi sensu lato prevalence in nymphal western black-legged ticks was 3.2% (95% CI = 1.9 to 5.0%) ( Table 2). The vast majority of Borrelia-positive nymphal ticks were collected in Marin County: 17/18 B. burgdorferi sensu lato-positive nymphs and 27/29 B. miyamotoi-positive nymphs were observed in Marin County (496 nymphs were collected in Marin County; a total of 155 nymphs were collected from other counties). All the B. burgdorferi sensu lato samples sequenced from nymphs were determined to be B. burgdorferi sensu stricto (10/10).
B. miyamotoi prevalence in nymphal ticks. In contrast to the adult stage, B. miyamotoi was observed in higher prevalence than B. burgdorferi in nymphal ticks at the regional level: B. miyamotoi prevalence was 5.1% (95% CI = 3.5 to 7.3%) ( Table 2), though there was no statistical difference (B. burgdorferi sensu lato versus B. miyamotoi; Fisher exact test P = 0.14; chi-square P = 0.10).
B. miyamotoi prevalence in larval ticks. A pooled sample of two larvae, collected at Olompali State Park, Marin County, tested positive for B. miyamotoi. All other larvae (n = 85) were negative for B. miyamotoi, and these were also collected in Marin County: Olompali State Park (n = 22), China Camp State Park (n = 16), Cascade Canyon Open Space (n = 35), Northern Marin (n = 9), Bolinas Lagoon (n = 2), and Samuel P. Taylor         Prevalence data are presented as number positive/number tested (percentage positive, 95% confidence interval). Superscripts indicate coinfections and represent one coinfected tick in the total sample. Sequenced samples are described following the reported prevalence, e.g., 6/90 ticks positive for Borrelia burgdorferi sensu lato, 2 of which were sequenced as B. burgdorferi sensu stricto. Only a subset of positive samples were sequenced.  Anaplasma phagocytophilum prevalence and coinfections. Anaplasma phagocytophilum was observed in Marin County at prevalence up to 7.8% (95% CI = 3.2 to 15.4%; n = 90; Bolinas Lagoon), though its presence was sporadic (4/14 sites), as well as in Monterey County (2/11 sites) and in Sonoma County (1/6 sites).
We observed three coinfected ticks: one adult tick from Salt Point State Park with B. burgdorferi sensu stricto (sequenced) and A. phagocytophilum, one nymph with a similar microbial combination (B. burgdorferi sensu lato-not sequenced) from Bolinas Lagoon, and one nymph with B. miyamotoi (sequenced) and A. phagocytophilum from China Camp State Park.
Coexisting tick-borne pathogens. There was no obvious pattern to coexistence of B. burgdorferi sensu lato and B. miyamotoi (Fig. 4). At sites where Borrelia species were observed, 13 sites had only B. burgdorferi sensu lato, six sites had only B. miyamotoi, and the two species coexisted at 10 sites.

Tick-Borne Pathogens in California
Applied and Environmental Microbiology At the site level, B. burgdorferi sensu lato was sometimes present in adult tick samples but absent from nymphs (e.g., China Camp State Park). The contrasting pattern (infected nymphs and uninfected adults) also appeared (e.g., Samuel P. Taylor State Park). Similarly, B. miyamotoi was observed in adult ticks at Samuel P. Taylor State Park but was not found in the collected nymphal ticks.
Habitat associations. Borrelia-positive ticks were observed in coastal chaparral and prairie habitats in Sonoma, Marin, Santa Cruz, and Monterey counties. Species identified by sequencing in these habitats represented the full gamut of Borrelia species identified in this study: B. burgdorferi sensu stricto, B. miyamotoi, B. americana, and B. bissettiae (Table 3; Fig. 3).
We compared tick-borne-pathogen prevalence in woodland and coastal chaparral habitat for our sample sites in Marin and Sonoma counties, using adult tick populations where n was .30. We restricted analyses to these two counties because they are coastal and the two habitats were well represented (n = 6 for woodland and chaparral sites), and we used just adult ticks because nymphs are difficult to collect in chaparral. There was no significant difference in the prevalence of B. burgdorferi sensu lato in the two habitats after aggregation of the data (x 2 = 0.03; P = 0.86), and site-level prevalence was similar (Fig. 5). In contrast, B. miyamotoi prevalence was higher in woodland habitats (x 2 = 5.57; P = 0.018) (Fig. 5). Prevalence of A. phagocytophilum did not differ according to habitat, though prevalence was low across sites (Fisher's exact test, P = 0.67).

DISCUSSION
Describing infection prevalence. The tick-borne pathogens Borrelia burgdorferi sensu lato, B. miyamotoi, and Anaplasma phagocytophilum were observed sporadically in questing tick populations across northern California. Aggregating across the region, we found higher infection prevalence of B. miyamotoi in nymphal ticks and higher B. burgdorferi sensu lato infection prevalence in adult ticks than reported in recent studies (12,26). Because the sampled ticks were collected at different times, these differences in prevalence may reflect trends in Borrelia infection patterns, interannual fluctuations in prevalence, or simply variation due to chance. Nevertheless, multiple measures of tick-borne infection prevalence are useful to gain a broader picture of local and regional pathogen prevalence, rather than relying on a single data source.
A key component and rationale of surveillance of ticks and tick-borne pathogens in the wild is to be able to represent the risk of exposure to humans. Often the data are summarized across the largest spatial extent. So, for example, despite sampling at multiple sites, data are reported as "In total, x% of ticks were infected with B. burgdorferi. . .," a behavior we have been guilty of (e.g., see references 17 and 29). However, aggregation of data into a single statistic can underrepresent the risk of exposure to tick-borne pathogens at sites where prevalence is higher, in part because the portrayed prevalence is deflated by including data from sites where tick abundance and/or pathogen prevalence is low. As an illustration, B. burgdorferi sensu stricto is known to be extremely rare in southern California-documented in only a single I. pacificus tick, from 5,571 ticks screened during three different studies (0.02%; 95% CI = 0.0005 to 0.1) (26,27,30). Consequently, describing B. burgdorferi prevalence in ticks for the state of California could dramatically underrepresent Lyme disease risk for northern Californians if all data are aggregated. This phenomenon can occur at a smaller scale: for example, B. miyamotoi prevalence in nymphal I. pacificus in Bolinas Lagoon, Marin County, was measured as 17.8%, with a 95% CI of 10.5 to 27.3% and a decent sample size of 90 nymphs. However, when aggregated across the county, nymphal infection prevalence of B. miyamotoi falls to 6.6% (95% CI = 4.4 to 9.4%), and in the Bay Area region, it slips to 5.1% (95% CI = 3.5 to 7.3) (Fig. 6; Table 3). Counties are often used as the spatial unit for reporting vectorborne and other disease metrics, but doing so can obfuscate smaller-scale patterns of disease risk (31) or result in erroneous interpretations of disease drivers (32). One solution to portray disease prevalence is to portray the 95% confidence intervals, which inherently demonstrate the range of interpretable prevalence. However, this method also has shortfalls if samples are aggregated, as the confidence intervals shrink with increasing sample size (Fig. 6), suggesting improved confidence but ignoring the fact that the source data are combined from multiple sites.
We advocate for transparently sharing data from all sites so that scientists, concerned citizens, physicians, public health agencies, and vector control districts can make appropriate judgments regarding the relevant risk of tick-borne disease. For example, it is important to understand that outdoor recreation in southern California has a lower risk for tick-borne disease exposure than outdoor recreation in northern California. These nuances can be important for treatment, control, and educational opportunities. In addition, zoonotic disease systems often exhibit fine-scale spatial patterns, and sharing these data at the site level may help future studies examining disease ecology and environmental drivers (33). Similarly, prevalence patterns of Borrelia likely will vary across time even at the same sites (34,35).
Typically, infection prevalence is reported for a single pathogen, e.g., prevalence of B. miyamotoi in a tick population or sample. This method of data presentation fails to recognize the fact that a population of ticks can often harbor multiple pathogens (35) and that reporting on a single pathogen species underestimates local risk of tick-borne disease. To provide an example, for the same Bolinas Lagoon tick population, pathogen prevalence is 6.7% for B. burgdorferi sensu lato, 17.8% for B. miyamotoi, and 7.8% for A. phagocytophilum. The overall prevalence of ticks with human pathogens in this population is 31.1% (28/90, as one tick was coinfected; 95% CI = 21.8 to 41.7%). The difference when multiple pathogens are considered is not always so pronounced; e.g., cumulative tick-borne-pathogen prevalence in China Camp State Park is 9.0% (6/61; 95% CI = 3.4 to 18.5%), compared to 7.5% for B. miyamotoi and 3.0% for A. phagocytophilum (one tick was coinfected). However, it is important to consider multiple pathogens when assessing local disease risk.
Vertical transmission of B. miyamotoi. B. miyamotoi is known to be vertically transmitted in Ixodes scapularis (18,19), has been observed in I. pacificus larvae (36), and is able to infect small mammals that ticks feed on (18,25,36,37). It is unclear whether infection dynamics in natural populations require amplification by horizontal transmission from the vertebrate hosts.
Recently, data from the Bay Area were used to argue that horizontal transmission is required for B. miyamotoi transmission in California, based on an increase in infection prevalence across developing tick life stages (36). However, this pattern was generated from data that included a single infected larva and aggregation of infection prevalence in tick life stages from eight different sites spanning five counties. At the site where the B. miyamotoi-infected larva was observed (Heinz Open Space, Santa Clara County), infection prevalence was 0.5% (1/201) in larvae, 0% in nymphs (0/19), and 0% (0/1) in adults (Fisher's exact test, P = 1.0). At sites with higher B. miyamotoi prevalence, e.g., Windy Hill, San Mateo County, there were significant differences between the life stages (Fisher's exact test, P = 0.01), though this statistical difference is driven by the lack of observed infected larvae (0/58 larvae, 5/57 nymphs, 16/137 adults; Fisher's exact test for only nymph and adult stages, P = 0.62).
In our study, at Olompali State Park, B. miyamotoi was observed in all I. pacificus life stages, though in only a single larva, and there was no significant change in infection prevalence. Because we also identified only a single B. miyamotoi-infected larva, interpretations of both studies on B. miyamotoi transmission across tick life stages should be viewed with caution. Though we suspect that small mammals do indeed play a role in B. miyamotoi infection dynamics, there are not yet enough field data from the California system to support this hypothesis. Increased surveillance for B. miyamotoi in larval I. pacificus and experimental tests of reservoir competence for B. miyamotoi in vertebrate hosts are required to demonstrate that horizontal transmission is important in the California system (25).
Borrelia ecology and habitat type. We observed a diversity of Borrelia species in coastal habitats. Coastal prairie and coastal chaparral have received relatively little attention compared to woodland habitats in northern California (e.g., see references 3, 17, and 38), and at first glance these habitats would appear to be low risk for Borrelia exposure due to the lack of recognized mammalian reservoir hosts; e.g., western gray squirrels are not common in these habitats. However, the prevalence of B. burgdorferi sensu lato in adult ticks in coastal chaparral in Marin and Sonoma counties was equivalent to that in woodlands, suggesting that this habitat may pose a risk for Lyme borreliosis exposure when adult tick populations are abundant in the winter.
Nymphal I. pacificus ticks were not collected in the coastal grass-or shrublands. We suspect that they are present but that tick flagging is not an effective way to collect this life stage in chaparral or grassland. Future investigations should attempt to survey tick hosts, e.g., western fence lizards, to examine the ecology of nymphal I. pacificus in these habitats (30).
Multiple Borrelia species (B. bissettiae, B. americana, Borrelia californiensis, and B. burgdorferi sensu stricto) have also been observed in coastal habitats in southern California (27) (Fig. 7; Table 1). Wood rats (Neotoma spp.) may play a role in Borrelia transmission in these environments, as they have been found to be infected with B. bissettiae, B. miyamotoi, and B. burgdorferi (21,22,25). Peromyscus mice may also be important in these habitats (25,27). Verification of host reservoir roles in coastal habitats requires further investigation, but the existing data suggest that Borrelia transmission dynamics are very different from the archetypal black oak woodland study systems, where wood rats and mice are believed to play largely peripheral roles in Lyme disease ecology (3,5).
B. americana was observed in three I. pacificus ticks, all collected in coastal chaparral habitat in Marin County. Prior observations of B. americana in I. pacificus were also linked to chaparral/grassland habitat in San Mateo and Los Angeles counties (26,30). In southern California, B. americana has also been observed in I. spinipalpis (26,27). Though data are still admittedly sparse, B. americana has been consistently observed in grassland/chaparral habitat, presumably because its reservoir host is associated with this habitat. Human infections with B. americana have not been reported (39).
A single B. bissettiae-infected tick was recovered from Monterey County. B. bissettiae has been associated with wood rats (Neotoma spp.) and other small mammals (Table 1) Table 1 for details and references). The maps were created in ArcMap, and the polygon feature class of the California county boundaries was downloaded from ArcGIS (credits: U.S. Bureau of Reclamation, California Department of Conservation, California Department of Fish and Game, California Department of Forestry and Fire Protection, and National Oceanic and Atmospheric Administration). and is potentially also a zoonotic pathogen, as it has been found infecting humans in northern California (20). It appears to be widely distributed in California's coastal region (Fig. 7).
We did not observe a pattern of dominance by either B. burgdorferi or B. miyamotoi, echoing previous reports from both California and the northeastern United States (13,40). Furthermore, based on the phylogeny, we found no evidence of geographic clustering of B. burgdorferi by latitude or sampling location ( Fig. 3 and 4).
Surveillance of adult versus nymphal ticks. Despite our best efforts, we struggled to find nymphs in Monterey County with tick flagging. However, adult I. pacificus ticks are abundant in Monterey County, and a variety of tick-borne pathogens are present (B. burgdorferi sensu stricto, B. miyamotoi, B. bissettiae, and A. phagocytophilum). Tick flagging is regarded as a sampling method that is representative of human exposure to ticks, and if this is the case, then human exposure to nymphs is rare in Monterey County Indeed, patterns of nymphal tick submissions from citizen scientists were rare in Monterey County (and counties further south) and were seen only in May (which is when we carried out surveillance for this study) (41). In contrast, citizen scientists reported adult ticks from Monterey County for several months (and from a broader swath of California) (41).
Although nymphs are regarded as the life stage that is most responsible for Lyme disease transmission (42), adult ticks are often easier to collect in abundance due to their habit of questing on higher vegetation and because they are more noticeable on tick flags. As such, adult ticks are good sentinels to demonstrate the local presence and diversity of Borrelia species. We observed B. bissettiae and B. americana only in adult ticks, though this may have been due to the larger samples as well as the habitat associations that appear to be important for B. americana ecology; i.e., it is difficult to collect nymphs in grassland/chaparral. Given the opportunity, we recommend that both adult and nymphal stages be included in tick-borne disease surveillance in California.

MATERIALS AND METHODS
Field sites and tick collection. Sampling sites were predominantly recreational areas or hiking trails, e.g., California state parks (SP) and midpeninsula open space preserves (OSP), in Marin, Mendocino, Monterey, Napa, Santa Clara, Santa Cruz, and Sonoma counties in northwest California (Table 3; Fig. 2). Some privately owned sites were also surveyed. Data are presented as belonging to a particular site which represents a single trail.
The study was conducted between December 2015 and May 2018 (Table 3; see also Data Set S1 in the supplemental material). Adult western black-legged ticks were predominantly collected each winter (December and January), when they are questing in greatest abundance (43). Collections in spring (May) were focused on nymphal ticks, though adults and larvae were also present and collected opportunistically. We attempted to visit each site during both winter and spring, but heavy rains and/or damage from wildfires precluded repeat visits in many locales. Olompali State Park was visited on three occasions.
Ticks were collected by dragging a 1-m 2 white flannel blanket along vegetation abutting trails for 20 m; ticks that attached themselves to the flannel were removed. We also collected ticks that were observed on vegetation, as well as any ticks found crawling on clothes or skin. We recorded the GPS coordinates and habitat type for observed ticks-either at the point that the ticks were observed or when a 20-m drag was successful in finding a tick. To prevent pseudoreplication of geographic data, we discarded GPS coordinates within 1.415 km of each other (44), unless the observed tick was a different life stage recorded in a different sampling period. Habitat classifications were coarse and included (i) coastal scrub/chaparral, where dominant species are coyote brush (Baccharis pilularis), California sagebrush (Artemisia californica), coastal buckwheat (Eriogonum parvifolium), sawtooth goldenbush (Hazardia squarrosa), and poison oak (Toxicodendron diversilobum); (ii) coastal grassland/prairie, which is dominated by annual grasses and forbs, with various amounts of native perennials; (iii) redwood forest, where dominant species are coastal redwood (Sequoia sempervirens) with associated Douglas fir (Pseudotsuga menziesii) and tanoak (Lithocarpus densiflorus); and (iv) oak-bay forest, where dominant species are coast live oak (Quercus agrifolia) or other Quercus species, California bay (Umbellularia californica), madrone (Arbutus menziesii), California blackberry (Rubus ursinus), and poison oak. At some sites, the trailside habitat was mixed, normally combining patches of coastal chaparral and grassland that could not be separated.
Ticks were stored in 70% ethanol. All ticks were identified to species and stage levels via morphology, and here we describe only observations of ticks identified as I. pacificus. DNA was extracted from ticks following manufacturer's protocols (DNeasy blood and tissue kit; Qiagen, Valencia, CA) and stored at 220°C until molecular analysis.
Pathogen detection and identification. To detect Borrelia pathogens, we used real-time PCR protocols described previously (17). In brief, we amplified a segment of the 16S rRNA gene of Borrelia sp. DNA (13), which enabled detection and classification of B. burgdorferi sensu lato (Lyme disease group) and B. miyamotoi (tick-borne relapsing fever group) through the detection of separate hybridization probes. Samples were considered positive if they had a cycle threshold (C T ) value of ,40 and logarithmic distributions on the amplification plots.
To identify Borrelia species and strain genotypes, we amplified and sequenced the 16S-23S intergenic spacer (IGS; rrs-rrlA) of a subset of the real-time PCR-positive tick samples using a nested-PCR protocol with a 25-ml reaction volume (45). The subset of Borrelia-positive ticks was chosen to represent as many different sites as possible across the geographical range of sampling. Prior to amplification of the inner target region, we used a 1Â magnetic bead cleanup to purify and concentrate the target DNA. During this magnetic bead cleanup, targets were annealed to the beads, washed twice with 70% ethanol (EtOH) and diluted into 12.5 ml of molecular-grade H 2 O before being added to the inner PCR mixture. Amplified samples were sequenced using capillary Sanger sequencing on an ABI 3730 sequencer with both forward and reverse primers (EnGGen, Northern Arizona University). Successfully sequenced forward and reverse Borrelia sp. samples were trimmed, and forward/reverse reads were assembled using Geneious prime (version 2019.1.1). For phylogenic reconstruction, sequences from this study were chosen from each location and were aligned (muscle alignment using default settings) with sequences obtained from GenBank NCBI (HQ012505.1 [46], KC416410.1 [47], EU886969.1 [48], EU377803.1, and EU377801.1 [49]) using MEGAX (version 10.1.8). A phylogenic tree was constructed with MEGAX using the maximum-likelihood method and the Tamura-Nei model (50,51).
Anaplasma phagocytophilum was detected using a previously described real-time PCR assay (52). We did not screen all ticks for A. phagocytophilum, so sample sizes differ from those for Borrelia sp. screening.
Analyses. Prevalence is reported as the percentage of ticks testing positive for the disease agent (i.e., number of positives/number tested Â 100). Some analyses were restricted to sites where sample sizes were .30, as this removes the impact of considering sites with an inflated pathogen prevalence because a single positive was observed in a small sample, and this seems to be an informal threshold at which we are normally able to detect Borrelia if it is present (16).
Binomial proportion 95% confidence intervals were calculated using binom.test in R. We used Fisher's exact test or the chi-square test to evaluate differences among proportions (i.e., infection prevalence).
Data availability. Sanger sequencing data have been uploaded in the NCBI database under accession numbers MW862414 to MW862434.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, XLSX file, 0.3 MB.