Borrelia burgdorferi Promotes the Establishment of Babesia microti in the Northeastern United States

Babesia microti and Borrelia burgdorferi, the respective causative agents of human babesiosis and Lyme disease, are maintained in their enzootic cycles by the blacklegged tick (Ixodes scapularis) and use the white-footed mouse (Peromyscus leucopus) as primary reservoir host. The geographic range of both pathogens has expanded in the United States, but the spread of babesiosis has lagged behind that of Lyme disease. Several studies have estimated the basic reproduction number (R 0) for B. microti to be below the threshold for persistence (<1), a finding that is inconsistent with the persistence and geographic expansion of this pathogen. We tested the hypothesis that host coinfection with B. burgdorferi increases the likelihood of B. microti transmission and establishment in new areas. We fed I. scapularis larva on P. leucopus mice that had been infected in the laboratory with B. microti and/or B. burgdorferi. We observed that coinfection in mice increases the frequency of B. microti infected ticks. To identify the ecological variables that would increase the probability of B. microti establishment in the field, we integrated our laboratory data with field data on tick burden and feeding activity in an R 0 model. Our model predicts that high prevalence of B. burgdorferi infected mice lowers the ecological threshold for B. microti establishment, especially at sites where larval burden on P. leucopus is lower and where larvae feed simultaneously or soon after nymphs infect mice, when most of the transmission enhancement due to coinfection occurs. Our studies suggest that B. burgdorferi contributes to the emergence and expansion of B. microti and provides a model to predict the ecological factors that are sufficient for emergence of B. microti in the wild.

emergence and expansion of B. microti and provides a model to predict the ecological factors that are sufficient for emergence of B. microti in the wild.
Although B. burgdorferi and B. microti are transmitted by the same vector in the northeastern and upper midwestern regions of the United States, the geographic spread of babesiosis has lagged behind that of Lyme disease [25,[31][32][33][34][35]. The delayed expansion of B. microti has been attributed to a lower efficiency of transmission between Peromyscus leucopus (white-footed mouse) and ticks [30] and to a narrower range of vertebrate reservoir hosts when compared with B. burgdorferi [36]. These observations are consistent with the lower basic reproduction number (R 0 ) reported for B. microti compared to that of B. burgdorferi. In fact, the B. microti R 0 has been estimated to be lower than the threshold for pathogen persistence (,1), raising the question of how it persists and expands in the northeastern United States [37,38].
Given that the establishment of B. burgdorferi typically precedes that of B. microti, we tested the hypothesis that coinfection of hosts with B. burgdorferi enhances the likelihood of B. microti establishment. To do so, we assessed the effect of coinfection at the individual host level in a laboratory setting that replicates pathogen-tick-host interactions that exist in the field. We then extended this observation to the population level by use of a mathematical model and identified ecological thresholds for B. microti establishment (R 0 .1). This model was made ecologically realistic by using data on ecological parameters obtained from two field sites in southern New England that are epidemiologically and ecologically distinct.

Laboratory infection experiments
Sources of mice and ticks P. leucopus mice (LL stock) were obtained from the University of South Carolina Peromyscus Genetic Stock Center and housed in a Yale Animal Resource Center facility. All procedures were approved by the Yale Institutional Animal Care and Use Committee (Protocol #07689). Mice were exposed to a diurnal light-dark cycle (16L:8D) and singly housed on wire cage bottoms over water to allow for collection of replete ticks. Mice were anesthetized prior to each infestation with nymphal or larval ticks. Infected I. scapularis nymphs were produced by allowing uninfected I. scapularis larvae to feed to repletion on infected mice. Fed larvae were collected and maintained in environmental chambers set at 21˚C and .90% relative humidity. After molting, nymphs were stored at 8˚C and in .90% relative humidity until experimental infestations. Uninfected I. scapularis larvae used for xenodiagnoses were produced by feeding wild-collected adult female I. scapularis on New Zealand White rabbits (Charles River Laboratories, Inc.); replete females were stored at 8˚C and in .90% relative humidity until they laid eggs, and moved to 21˚C and in .90% relative humidity until the hatching of eggs.
Infection of P. leucopus mice with B. microti and B. burgdorferi The experiment was carried out in two sets. In the first set ( Fig. 1), P. leucopus mice were infested with nymphal ticks infected with B. microti, or a combination of nymphal ticks infected with B. microti and either the B. burgdorferi strain BL206 or the B. burgdorferi strain B348. The B. microti strain was previously isolated from a Peromyscus leucopus mouse trapped in Greenwich, CT and maintained by alternate passaging between C3H/HeJ Prkdc scid mice and I. scapularis ticks [31]. The two B. burgdorferi strains have polarized infectious phenotypes. BL206 is characterized by an Osp C genotype A and is highly invasive; it migrates from the skin into the bloodstream and reaches secondary sites. In contrast, B348 is characterized by an Osp C genotype E and is non-invasive as it remains at the tick bite site [39][40][41][42][43][44][45][46]. In the second set of experiments, mice were infected with B. microti alone or together with B. microti and B348. Data from these two sets were combined for statistical analyses.
Assessment of B. microti transmission from P. leucopus mice to ticks We used xenodiagnoses to assess pathogen transmission from mice to ticks, as previously described [47,48]. One hundred uninfected larval ticks were placed on each mouse on days 7, 14, 21, 28 and 42 post nymphal infestation. Fed larvae were collected from water trays placed beneath the mouse cages and maintained in environmental chambers until they molted into nymphs. The B. microti burden was assessed in 20 nymphs that were randomly selected among those obtained from each mouse.

Detection of B. microti DNA in ticks
Individual nymphal ticks that had been stored frozen in liquid nitrogen were homogenized using sterile pestles. DNA was extracted using the DNeasy Blood and Tissue Kit (QIAGEN, Valencia CA), and eluted in 120 mL of 10 mM Tris?HCl at pH 8.5. The B. microti 18S rRNA gene (GenBank accession number AY144696.1) was amplified by quantitative PCR [49] using these forward and reverse primers and probe (from 59 to 39): AACAGGCATTCGCCTTGAAT, CCAACTGCTCCTATTAACCATTACTCT, and 6FAM-CTACAGCATGGAATAATGA-MGBNFQ, respectively. The PCR reaction consisted of 2X Taqman Universal PCR Master Mix (with AmpErase, Applied Biosystems, Foster City CA), 0.9 mM forward and reverse primers, 0.2 mM probe, and 5 mL DNA template in a total reaction volume of 25 mL. DNA was amplified in an Applied Biosystems 7500. Ticks were considered positive for B. microti DNA if amplicons were detected at or below a cycle threshold (C T ) value of 35 [49].

Assessment of B. microti parasitemia in mice
On days 7, 14, 28, and 42 after nymphal infestation of P. leucopus mice, 50 mL of peripheral blood were obtained via submandibular venipuncture. Blood cells were fixed in glutaraldehyde, permeabilized in Triton X-100, and treated with DNasefree RNase, as previously described [50]. Parasites were detected using the nucleic acid dye YOYO-1 iodide. For each mouse blood sample, 10,000 cells were acquired using the FACS Calibur (Becton Dickinson, San Jose, CA). Upon excitation by the argon-ion laser (488 nm), fluorescence emission (509 nm) was recorded using CellQuest (Scripps Research Institute, La Jolla, CA) and analyzed using FlowJo (TreeStar, Ashland, OR).

Statistical analyses
We examined whether coinfection of mice with B. microti and B. burgdorferi increases the probability of nymphal infection with B. microti, when compared with mice infected with B. microti alone. Statistical analyses were performed using generalized estimating equation population-averaged logit models in STATA/SE, version 12.0 (STATA Corporation, College Station, TX). These generalized linear models allow specification of the within-group (panel) correlation structure. Because multiple ticks were allowed to feed on an individual mouse, a within mouse correlation structure was specified in the model to account for the autocorrelation in the infection status among the ticks that fed on the same mouse. We treated time as a continuous variable (days since infection with B. microti and/or B. burgdorferi) and assessed interactions between mouse group and time. Differences in parasitemia among groups were assessed using negative binomial regression, including a term to account for days since infection with B. microti.

Field sampling
To obtain realistic estimates of ecological parameters as inputs for modeling of B. microti emergence, we determined the tick burdens and the seasonal pattern of tick feeding (phenology) on P. leucopus mice trapped at two ecologically distinct sites: eastern Connecticut (Nehantic and Pachaug State Forests) and Block Island, Rhode Island (Old Mill, Rodman's Hollow, and West Beach). The host community on Block Island is dominated by P. leucopus and is less diverse than that at the Connecticut sites. Furthermore, the tick burdens are higher at the Block Island sites, when compared with the mainland sites [51]. . At every site, trapping occurred bimonthly for three consecutive nights per session for a total of seven sessions at the Connecticut sites and six sessions at the Block Island sites. P. leucopus mice were live-trapped in Sherman box traps (9063063.50) which were baited with rolled oats, sunflower seeds and cotton balls, set at dusk, and checked shortly after dawn the next day. Captured individuals were ear-tagged, aged, sexed, weighed, searched for ticks, bled by submandibular venipuncture, and subjected to ear punch biopsy before release at the point of capture. Ticks were removed with forceps and preserved in 70% ethanol. All field study procedures were approved by the Yale Institutional Animal Care and Use Committee (Protocol #07596). Inhalation of isoflurane was the approved method of euthanasia for situations when warranted, however, no animals had to be euthanized during this study. Field studies did not involve endangered or protected species. Property access permissions and scientific collector's permits were obtained from the Connecticut Department of Energy and Environmental Protection, the Rhode Island Department of Environmental Management, The Nature Conservancy, and the US Fish & Wildlife Service (Charlestown, RI).

Mathematical (R 0 ) modeling
To determine whether host coinfection significantly increases the likelihood of B. microti establishment in the wild, we integrated laboratory and field data into a mathematical model that estimates the basic reproduction number R 0 . The basic reproduction number describes the expected initial spread of a pathogen that arrives in a naïve host population and is used to predict the ecological conditions that allow pathogen establishment (parameter ranges for which R 0 .1) or result in pathogen fade-out (parameter ranges for which R 0 ,1) [52]. Larger R 0 values also imply an increased likelihood of pathogen establishment and therefore a shorter time to establishment. Our model is derived from the model constructed by Dunn et al. (2013) [38], which was based on the original model by Hartemink et al. (2008) [53] and modified by Davis and Bent (2011) [37]. These models assume that R 0 is a function of one B. microti infected nymph or mouse introduced into an already established B. burgdorferi infected population. Parameters used in the R 0 calculation were estimated from our laboratory and field experiments (see above) and from the literature. Parameter definitions and parameter point values for the Connecticut and Block Island sites are shown in Table 1 [37,38,[54][55][56][57][58][59][60][61]. Based on serological studies of mammals trapped in the wild, we considered two prevalence rates for B. burgdorferi infection in P. leucopus mice: a high rate of 0.80 and a low rate of 0.30 [12,62,63]. Additional information, including full parameter ranges and the results of a global sensitivity analysis of R 0 , is published elsewhere [38]. Based on evidence of sequential geographic expansion of B. burgdorferi and B. microti [25,34], our model assumes that B. microti is spreading into new sites where B. burgdorferi is already enzootic. The formulation assumes that a constant fraction (c) of P. leucopus mice is infected with B. burgdorferi (Fig. 2), thereby defining three types of hosts/vectors: (1) a P. leucopus infected with B. microti, (2) a P. leucopus coinfected with B. burgdorferi and B. microti and, (3) a tick infected during its first blood meal. The model was restricted to P. leucopus mice because they are the main vertebrate hosts for B. microti [36] and because laboratory transmission efficiency data are available for this species; however, our modeling does take into account that a proportion of the tick population, 1-c, feeds on other hosts ( Table 1).
The next-generation matrix is constructed according to the transmission probabilities in Fig. 2 with the k ij defined as the expected number of host type i infected by a single individual of host type j over its entire infectious period (a A full description of parameters and corresponding references and field data can be found in [38]. Burden parameters assume January 1 st as day 0. doi:10.1371/journal.pone.0115494.t001 zero entry indicates transmission from host type j to host type i is negligible). R 0 is then the dominant eigenvalue of the next-generation matrix (Fig. 2). The functional forms of k 23 , k 32 , k 13 and k 31 are adapted from [53] where k 23 and k 13 are, respectively, the expected number of uninfected P. leucopus infected with B. microti from a nymphal tick infected with B. microti, and the expected number of B. burgdorferi-infected P. leucopus infected with B. microti from a nymphal tick infected with B. microti. R 0 takes the form Here, s N , q N , c and d L , respectively, represent the probability of survival from infected fed larva to infectious feeding nymph, the probability of transmission from nymph to mouse, the proportion of ticks feeding on P. leucopus, and the duration of tick attachment for larvae taking a blood meal. The parameter h takes into account mouse survivorship with lifespan r and is given by, The functions Z L t ð Þ and a N (t) are, respectively, the mean larvae burden and the scaled mean nymphal burden on P. leucopus and are given by, and, The phenology curves, Z L t ð Þ and Z N t ð Þ, have well defined parameters for the heights of each peak, timing of beginning of activity and time to peak activity (S1 Fig.). There is an additional shape parameter which controls the tail of the distributions. These curves were fitted to our field data consisting of larval and nymph counts from trapped mice. The data were pooled for the two sites in Connecticut and the three sites on Block Island, Rhode Island using the same combinations of normal and right-shifted log-normal functional forms as previously published [60]. These curves represent the phenology of the immature life stages of I. scapularis as observed in areas of the northeastern and upper midwestern United States where both B. burgdorferi and B. microti are endemic. The functional forms used are informed by a large number of field studies that show an initial normal peak in questing larvae in late spring followed by a larger second log-normal peak. Nymphal activity is well represented by a single, lognormal curve [60]. Finally, the efficiency of transmission of B. microti from mice singly infected with B. microti and from mice coinfected with B. microti and B. burgdorferi, as measured from the laboratory experiments (see Results), are included by constructing functions, p 1 (t) and p 2 (t). These are piecewise linear functions fitted to the transmission efficiencies observed at 7, 14, 21, 28 and 42 days after infection. Transmission efficiency is assumed to simply change linearly between these observed values. The 95% confidence intervals were constructed by using bootstrap methods [64].

Results
Pathogen interactions in the laboratory setting: the frequency of B. microti infected nymphs increases when they feed as larvae on P. leucopus mice coinfected with B. burgdorferi We determined the effect of B. burgdorferi host coinfection on the acquisition of B. microti by ticks. The frequency of B. microti infected nymphs was higher when larvae fed on mice coinfected with B. microti and the highly invasive B. burgdorferi strain BL206 (Group 2, Fig. 1) than on mice infected with B. microti alone (Group 1, Fig. 1) (odds ratio 53.73, p,0.05) (Fig. 3A, S1 Table). The frequency of B. microti infected nymphs declined over time following infection of the host, as indicated by a significant effect of days post-infestation on the percentage of infected ticks (odds ratio 50.95, p,0.001) (Fig. 3A, S1 Table).
Nymphs derived from larvae that had fed on mice coinfected with B. microti and the non-invasive B. burgdorferi strain B348 (Group 3, Fig. 1) were as frequently infected with B. microti as those derived from larvae that had fed on mice infected with B. microti alone (Group 1, Fig. 1 and Fig. 3B). Given that the frequency of B. microti infected ticks appeared to increase during the first two weeks after infection, we next compared the frequency of infection in ticks that had fed on coinfected and non-coinfected mice during this short time period. During the first 14 days post-infection, coinfection significantly increased the frequency of B. microti infected nymphs (odds ratio 54.6, p50.045).
We determined whether the higher frequency of B. microti-infected nymphs was associated with a higher B. microti parasitemia in mice. Parasitemia peaked on day 14 whether mice were coinfected with B348 and B. microti or with B. microti alone (Fig. 3B inset). Parasitemia was higher in the coinfected group than in the B. microti only group (negative binomial regression, LR Chi2 523.4, p,0.001), consistent with the higher frequency of B. microti-infected nymphs in the coinfected group. B. microti parasitemia was not monitored in mice coinfected with BL206.
Ecological parameters influencing pathogen transmission in the field: higher tick burdens on Block Island but greater synchrony of larval and nymphal feeding in Connecticut We examined factors that would enhance pathogen transmission at our study sites. We quantified the tick burdens on mice trapped at sites in eastern Connecticut and on Block Island, Rhode Island. Larval and nymphal tick burdens were higher on mice at the Block Island sites than on those trapped in Connecticut (Fig. 4, S2 Fig.). At both sites, nymphal activity peaked in late spring (June 9 in both Connecticut and Block Island). The first peak of larval activity also occurred in late spring (June 4 in Connecticut; June 13 on Block Island) and overlapped with nymphal activity whereas the second peak was reached in mid to late summer (August 19 in Connecticut; August 12 on Block Island) (Fig. 4, S2  Fig.). The more intense spring larval peak in Connecticut significantly overlapped with the nymphal activity peak and thus provided B. microti transmission opportunities during the early stages of infection when there was a stronger effect of coinfection on transmission efficiency (Fig. 3 and Fig. 4). In contrast, most larval activity occurred in late summer on Block Island, resulting in a small overlap with the tail end of the spring nymphal activity, i.e., when transmission efficiency is no longer enhanced by coinfection (Fig. 4).
Integration of laboratory-derived infection parameters with fieldderived tick burdens: co-infection of reservoir hosts with B. burgdorferi lowers the ecological requirements for B. microti establishment We integrated laboratory and field data into a mathematical model that estimates the basic reproduction number R 0 . Based on serological studies of mammals trapped in the wild, we considered two point prevalences for B. burgdorferi infection in P. leucopus mice: a high prevalence of 0.80 and a low prevalence of 0.30 [12,62,63]. R 0 values are presented as a function of the proportion of infected larvae that survive and molt to become feeding infectious nymphs (S N ) and of the proportion of larval ticks feeding on P. leucopus (c) (Fig. 5, S3 Fig.). These two parameters were chosen on the basis of previous work [38] that ranked them as two of the most important determinants of R 0 , although both were ranked below the parameters governing pathogen transmission efficiency that were measured directly in the present study [38]. The effect of co-infection with a disseminating strain of B. burgdorferi (BL206) is presented in Fig. 5 whereas the effect of co-infection with a non-disseminating strain (B348) is presented in S3 Fig. Curves were fitted to the experimental infection data (see Methods) by considering nymphs fed as larvae on co-infected mice (red curves) and those fed as larvae on singly infected mice (blue curves). The larger the area above the curve ('emergence'), the broader the conditions that allow for the establishment of B. microti. Taking into account 95% confidence intervals (calculated from repeated resampling of the data presented above and represented by dotted curves in Fig. 5), we concluded that coinfection significantly enhances the likelihood of B. microti establishment in Connecticut (increase in R 0 of 15%) and on Block Island (increase in R 0 of 11%) when B. burgdorferi prevalence among P. leucopus is high (0.80) (confidence bounds in Fig. 5 right panels do not overlap).
Lastly, we estimated the threshold of B. burgdorferi prevalence in P. leucopus above which R 0 values are significantly increased by coinfection as determined by no overlap between the 95% confidence intervals for all values of S N and c. This threshold was 0.42 for the Block Island sites and 0.25 for the Connecticut sites.

Discussion
We have demonstrated in the laboratory that the frequency of B. microti-infected ticks is higher when fed on P. leucopus that are coinfected with B. burgdorferi and B. microti than on mice infected with B. microti alone. In field studies, we quantified two important ecological parameters that affect pathogen transmission, tick burdens on P. leucopus and timing of tick feeding. By use of a mathematical model that integrates laboratory and field derived data, we identified ecological conditions that synergize with co-infection to enhance B. microti establishment (R 0 .1). Our modeling results indicate that high prevalence of B. burgdorferi in P. leucopus significantly lowers the ecological thresholds for enzootic establishment of B. microti at mainland and island sites, with a stronger effect at mainland sites. These results suggest that the geographic spread of B. microti is favored by prior enzootic establishment of B. burgdorferi.
The effects of coinfection on the I. scapularis-borne pathogen infection cycle vary according to pathogen species and genotype, as observed in previous studies of B. microti and of the interactions between B. burgdorferi and A. phagocytophilum [30,47,48]. The two B. burgdorferi strains used in this study had markedly different phenotypes in regard to host invasiveness and duration of infection. The proportion of B. microti-infected nymphs was significantly increased when mice were coinfected with the highly invasive B. burgdorferi strain BL206. This enhancement lasted over the initial four week study period. We also observed increased nymphal infection when larvae had fed on mice infected with both B. microti and the non-invasive B. burgdorferi strain B348, but this effect was restricted to the first two weeks of infection. Consistent with this observation, B. microti parasitemia was transiently higher in mice coinfected with the B348 strain than with B. microti alone. The highly invasive B. burgdorferi strain BL206 can be isolated from human blood and can be transmitted from mice to ticks for about 80 days post-infection [41,[44][45][46][65][66][67]. In contrast, the non-invasive B. burgdorferi B348 strain often is found in human skin without dissemination and is efficiently transmitted from mice to ticks for only approximately 40 days [41,[44][45][46]. The difference in dissemination between B. burgdorferi strains may contribute to the difference in their ability to promote B. microti transmission, as the immune response elicited by these strains likely differ in nature and possibly duration. By remaining in the dermis at the site of a tick bite, non-invasive B. burgdorferi strains may be poorly able to modulate the B. microti specific immune response that develops in the spleen [15,68]. In contrast, dissemination of a B. burgdorferi strain through the bloodstream may significantly impair the immune response required to control and eradicate B. microti in the host [15,[69][70][71]. Although our experiments were restricted to two representative strains, the B. burgdorferi strain phenotypes are likely to be relevant to P. leucopus infections in the field. Indeed, the monophyletic clade to which B. burgdorferi strain BL206 belongs [39] is highly prevalent throughout the northeastern United States [72] and strains that clustered with the BL206 strain are likely to share the invasive phenotype.
The increased proportion of B. microti-infected nymphs derived from larval feeding on B. burgdorferi coinfected mice may help explain geographic differences in endemicity and the pattern of emergence of human babesiosis. B. burgdorferi has expanded more rapidly than B. microti, presumably because it is transmitted more efficiently between ticks and white-footed mice [30]. Our observation that host coinfection increases the proportion of B. microti infected nymphs may help explain the high B. microti prevalence rates in some areas that have long been endemic for B. burgdorferi [25,33]. Our modeling results show that the larger tick burdens observed on P. leucopus on Block Island lower the predicted ecological thresholds for B. microti establishment. This is consistent with the high endemicity of B. microti on New England coastal islands. In contrast, lower tick burdens result in higher ecological thresholds for B. microti establishment at the Connecticut sites. These thresholds may be reduced by coinfection when a large proportion of white-footed mice are infected with B. burgdorferi. Thus, the delay in B. microti establishment in Connecticut and other mainland sites may be attributed to the time required for establishment of B. burgdorferi [73,74].
The effect of coinfection on B. microti establishment also depends on the timing of feeding by nymphs and larvae. I. scapularis-borne pathogens typically are maintained in their enzootic cycles by sequential transmission from infected nymphs to white-footed mice to uninfected larvae. A short time gap between the feeding by nymphs and the feeding by larvae increases the chance for completion of the enzootic cycle because the reservoir host is less likely to die or to mount an immune response that could eliminate the pathogen. In the northeastern United States, nymphs typically feed from late spring to early summer whereas larvae feed in two periods, one in late spring and the second one in late summer [75]. The magnitude and duration of the two larval feeding periods vary between and within endemic areas [72,76]. White-footed mice are more likely to become infected with B. burgdorferi and B. microti from late spring to early summer, i.e., when infected nymphs transmit both pathogens during their blood meal. Given that the B. microti parasitemia in white-footed mice is high and the transmission of B. microti to feeding larvae more likely during the first two weeks after infestation in experimental mice, we expect the transmission efficiency to feeding larvae in the wild to be highest during or soon after the first larval feeding period. In this context, the intense and prolonged larval questing activity seen in late spring in Connecticut may synergize the effect of coinfection on B. microti establishment as revealed by an increased R 0 . In contrast, the more limited impact of coinfection on B. microti establishment on Block Island is consistent with an intense larval feeding activity in late summer. On Block Island the effect of coinfection on B. microti transmission fades as the summer progresses but B. microti transmission most likely remains high because of the large tick burden of P. leucopus mice. Synchronous feeding previously has been shown to play a key role in the maintenance of flaviviruses causing tick-borne encephalitis in Europe [77] and in the United States [78]. We demonstrate that it also plays a key role in the maintenance and potential for interaction of pathogens with short infectious periods. Similar to our Connecticut study sites, spring larval activity is intense in the upper Midwest [72], thereby raising the likelihood of increased B. microti transmission by concurrent B. burgdorferi infection in vertebrate hosts.
In summary, we have observed increased B. microti transmission from B. burgdorferi coinfected P. leucopus mice to I. scapularis ticks in the laboratory. We found that the strength of the coinfection effect depends on the B. burgdorferi strain, the tick burden on the primary vertebrate host (P. leucopus), and the overlap between nymphal and larval feeding periods. We incorporated these factors in a mathematical model to predict establishment of B. microti in a region. The model predicted that coinfection enhancement is stronger in Connecticut than on Block Island, which may partly explain why B. microti has lagged behind B. burgdorferi establishment on the mainland. We are now in a position to identify ecologically suitable areas for future expansion of B. microti. Our findings also imply that control measures such as reservoir host vaccination against B. burgdorferi may reduce B. microti transmission and therefore viability in areas that are highly endemic for B. burgdorferi [79][80][81]. Lastly, our integration of experimental and field data into a realistic model of R 0 is a powerful approach to examine the effects of coinfection on other tick-borne pathogens as well as other pathogens transmitted by vectors other than ticks. . This figure shows differences in threshold curves, representing where R 0 51, and associated confidence intervals that separate regions of s N and c where B. microti is expected to emerge and regions where it is expected to fade out. Threshold curves are contour curves where R 0 is plotted as a function of two variables: the proportion of fed infected larvae that survive to become infectious feeding nymphs, s N , and the proportion of ticks feeding on Peromyscus leucopus, c. Plots indicate effects of location specific (Block Island and Connecticut) timing of tick activity as well as B. burgdorferi strain BL348 prevalence in mice (low 50.3 and high 50.8) on R 0 . Although the curves separate, the confidence intervals overlap, implying that coinfection with the B. burgdorferi strain B348 did not significantly change the expected value of R 0 . Differences in threshold curves for B. burgdorferi strain BL206 are shown in Fig. 5. doi:10.1371/journal.pone.0115494.s003 (PDF) S1