Host Contributions to the Force of Borrelia burgdorferi and Babesia microti Transmission Differ at Edges of and within a Small Habitat Patch

ABSTRACT

In the northeastern United States, the emergence of Lyme disease has been associated, in part, with the increase of small forest patches. Such disturbed habitat is exploited by generalist species, such as white-footed mice, which are considered the host with the greatest reservoir capacity for the agents of Lyme disease (Borrelia burgdorferi sensu stricto) and human babesiosis (Babesia microti). Spatial risk analyses have identified edge habitat as particularly risky. Using a retrotransposon-based quantitative PCR assay for host bloodmeal remnant identification, we directly measured whether the hosts upon which vector ticks fed differed at the edge or within the contiguous small habitat patch. Questing nymphal deer ticks, Ixodes dammini, the northern clade of Ixodes scapularis, were collected from either the edge or within a thicket on Nantucket Island over 3 transmission seasons and tested for evidence of infection as well as bloodmeal hosts. Tick bloodmeal hosts significantly differed by site as well as by year. Mice and deer were identified most often (49.9%), but shrews, rabbits, and birds were also common. Ticks from the edge fed on a greater diversity of hosts than those from the thicket. Surprisingly, mice were not strongly associated with either infection at either sampling site (odds ratio [OR] < 2 for all). Although shrews were not the most common host utilized by ticks, they were highly associated with both infections at both sites (OR = 4.5 and 11.0 for B. burgdorferi and 7.9 and 19.0 for B. microti at the edge and thicket, respectively). We conclude that reservoir hosts may differ in their contributions to infecting ticks between edge and contiguous vegetated patches.

IMPORTANCE Habitat fragmentation is thought to be a main factor in the emergence of Lyme disease and other deer tick-transmitted infections. The patchwork of forest and edges promotes altered biodiversity, favoring the abundance of generalist rodents, such as white footed mice, heretofore considered a key tick and reservoir host in the northeastern United States. We used tick bloodmeal analyses to directly identify the hosts from which nymphal deer ticks became infected. We demonstrate that there is considerable microfocality in host contributions to the cohort of infected ticks and that shrews, although they fed fewer ticks than mice, disproportionately influenced the force of pathogen transmission in our site. The venue of transmission of certain deer tick-transmitted agents may comprise a habitat scale of 10 m or fewer and depend on alternative small mammal hosts such as shrews.

INTRODUCTION

The emergence of Lyme disease and other tick-transmitted zoonoses in the northeastern United States has been attributed to changes in the landscape, suburbanization, and increased density of deer, the main reproductive host for the tick vectors, Ixodes dammini, the “northern” clade of Ixodes scapularis (1). Habitat fragmentation, in particular, contributes to great risk. Colonial activities removed at least half of the forest, and almost none remained uninfluenced by human activity (2). Much reforestation has taken place in the interim. The newly reforested areas differ in their composition and are considerably more fragmented due to development. Indeed, the amount of wildlife-urban interface where human development abuts public or private wildlands, as measured for wildfire risk, has increased by 52% since the 1970s (3). Smaller forest patches are associated with greater levels of apparent entomological risk, as measured by the density of infected nymphal ticks (4, 5). As habitat patch size diminishes, so does species richness (6), and habitat generalists predominate. Peromyscus leucopus, the white-footed mouse (hereafter mouse), is one such habitat generalist (7). These mice are widely considered the host with the greatest reservoir capacity for Borrelia burgdorferi sensu stricto and Babesia microti, the agents of Lyme disease and human babesiosis, in the northeastern United States. (8, 9). Subadult ticks are more likely to feed on highly reservoir competent hosts, such as mice, in small patches where other host species are less well represented, thereby promoting the force, or intensity, of B. burgdorferi or B. microti transmission (10). Humans living within a patchy landscape are thus at increased risk of tick-borne infection (11, 12).

Edges, defined as forest or herbaceous habitat transitions, have been identified in spatial risk analyses as areas that have increased entomological risk for tick-borne infections, termed the “edge effect” (13–15), although it is unclear whether this effect is due to entomological correlates (e.g., more ticks) or greater human exposure at forest edges. Small forest patches are characterized by their high proportion of edge habitat to overall area. Abrupt induced edges (16) that characterize peridomestic sites, as opposed to inherent edges (habitat transitions due to natural features, e.g., soil type) may be even more relevant to risk for tick-borne infection. Mice are often denser at edges in some but not all sites (17–19). Tick densities in endemic communities are greater in the nearby forest than at the edge (19–24), but ticks collected at edges were more likely to be infected than those collected in the forest (22). However, this association is not consistently reported (21, 25, 26) and in fact may differ between transmission seasons (27). Interestingly, sometimes fewer ticks infest mice from small patches than larger ones (28), which might suggest that transmission would be less intense in edge sites. Although species diversity in small patches tends to be lower than that for larger ones, many deer tick hosts, such as deer, chipmunks, shrews, rabbits, birds, raccoons, and opossums (29), effectively utilize small patch habitats. Hosts with greater reservoir capacity, such as mice, are not necessarily more prevalent in small as opposed to large patches (28, 30); chipmunks may be as common as mice (31, 32). Given the lack of consistency between studies, additional fine-grained analyses of the entomological correlates of the “edge effect” would be useful. If entomological risk is greatest at the edges of small patches, environmental interventions may be more narrowly focused.

The actual contribution of a species to enzootic B. burgdorferi transmission is difficult to quantify and has largely been based on syntheses of indirect measurements. Models extrapolate from a few measured parameters, such as host population density, tick infestation indices, tick developmental success, and relative host reservoir competence (33–37). Most of these parameters rely on trapping animals, the success of which is species dependent. Some animals, such as mice, are easily trappable and some, such as shrews, may not be, particularly using protocols optimized for trapping mice. Hence, site-specific indices of abundance may be more representative for some hosts compared to others. Direct identification of the host from which a questing tick became infected by means of host bloodmeal remnant analysis more precisely measures host contributions to the force of transmission. We have recently developed a highly sensitive assay based on PCR amplification of retrotransposons that is capable of determining whether host-seeking nymphal deer ticks had fed on a mouse or deer as larvae (38). This assay has now been extended to identify other likely hosts for subadult deer ticks in enzootic sites, such as shrews, rabbits, voles, squirrels, and birds (39). We use these assays to determine whether ticks at induced narrow habitat edges derive from different hosts than those in the adjacent dense vegetation. In particular, we sought to test the hypothesis that the representation of hosts contributing B. burgdorferi- or B. microti-infected ticks differed between the edge and adjacent habitat on a small spatial scale.

RESULTS

Overall, ticks collected at the edge were more likely to be infected with either B. burgdorferi or B. microti than those collected in the thicket (Table 1). Prevalence estimates from edge-collected ticks were greater than those of thicket ticks for both infections (20% versus 13% for B. burgdorferi, P = 0.07, and 16% versus 8% for B. microti, P = 0.01, for all years combined). The last collection year, 2020, was the exception; the prevalence of B. burgdorferi was equal from both sites. The number of ticks collected from the edge per minute of effort was less than in the thicket (overall 0.8 ticks per minute at the edge versus 1.3 in the thicket, P = 0.000) (Table 2). Once again, there was an exception; in 2018 equal numbers of ticks were collected per minute of effort. Because the edge yielded fewer ticks but more of them were infected, and, conversely, the thicket yielded more ticks but fewer of them were infected, the entomological risk index (also known as density of infected nymphs [DIN]) for both sites did not differ (Table 2) (10).

TABLE 1.

Prevalence of Borrelia burgdorferi and Babesia microti in deer ticks at the edge versus thicket

Site	Yr	No. tested	Borrelia burgdorferi		Babesia microti
			No. positive	% Positive (95% CI)	No. positive	% Positive (95% CI)
Edge	2018	65	14	22 (21–33)	10	15 (8–26)
	2019	88	20	23 (14–33)	15	17 (10–27)
	2020	61	8	13 (6–24)	10	16 (8–28)
	All	214	42	20 (15–26)a	35	16 (12–22)b
Thicket	2018	67	8	12 (5–22)	6	9 (3–18)
	2019	63	9	14 (7–25)	5	8 (3–18)
	2020	87	12	14 (7–23)	7	8 (3–16)
	All	217	29	13 (9–19)a	18	8 (5–13)b

^aP = 0.07.

^bP = 0.01.

TABLE 2.

Estimation of the entomological risk index^a

					Borrelia burgdorferi		Babesia microti
Site	Yr	No. collected in June	No. mins	No. ticks per min	No. positive	Risk index	No. positive	Risk index
Edge	2018	65	60	1.1	14	0.23	10	0.17
	2019	88	95	0.9b	20	0.21	15	0.16
	2020	61	120	0.7c	8	0.07	10	0.07
	All	214	275	0.8d	42	0.15	35	0.13
Thicket	2018	67	60	1.1	8	0.13	6	0.1
	2019	63	30	2.1b	9	0.3	5	0.17
	2020	66	60	1.1c	6	0.1	4	0.07
	All	196	150	1.3d	23	0.15	15	0.1

^aRisk index, number of infected ticks divided by the number of minutes of effort.

^bP = 0.002.

^cP = 0.01.

^dP = 0.000.

The success rate for bloodmeal identification over the course of the study was high. Overall, bloodmeal host was identified from 83% of the ticks collected, and 14.0% tested positive for more than 1 host, even after the assumed external contamination was removed from the data set. Mice and deer were responsible for feeding the most ticks (49.9%), but host utilization was highly variable; different species dominated each year and at each site (see Fig. S1 to S7 in the supplemental material). For example, at the edge in 2018, mice, shrews, and rabbits (all greater than 30% of ticks) were more common than deer (7.6%) (Fig. S1), but in 2019, deer were more common than the others (30.7% deer versus <20% each for shrews, mice, and rabbits) (Fig. S3). In 2020, shrew (24.5%), mouse (18%), and bird (13%) were the most common; deer contributed only 9.8% of the bloodmeals. In the thicket, deer and mice remained dominant in 2018 and 2019 (over 30% for each) (Fig. S1 and S3). Mouse contributions dropped in 2020 to 16%, but deer contributions remained high (33%).

Ticks collected from the edge were significantly more likely to have fed on rabbits and shrews (P = 0.000) (Fig. 1) than those collected from the thicket, and overall, the diversity of bloodmeal hosts was higher from edge-collected ticks (Table 3). Ticks from the thicket were significantly more likely to have fed on deer (P = 0.000) or have an unidentified bloodmeal (P = 0.005).

FIG 1.

Bloodmeal analysis of nymphal deer ticks collected at the edge (A) versus the thicket (B) using the full data set with external contamination removed. The data are presented as a percent of the total number of ticks that tested positive at each site for each host type; all years are combined. The colors of the bars indicate the other species identified in multiple bloodmeals. Asterisks indicate that the percent of ticks with that host type is significantly greater at one site versus the other (Fisher’s exact test, P ≤ 0.05). Multi, ticks with multiple bloodmeals detected; sqrl, squirrel; TBD, to be determined.

TABLE 3.

Comparison of the diversity of species feeding ticks at each site^a

Site	Shannon index
Edge	1.95 (1.88–1.99)
Thicket	1.77 (1.67–1.84)

^aShannon index of diversity with 95% confidence intervals are reported for the full data with contaminating bloodmeals removed (P value of 0.000 between the two sites).

Because larval ticks emerge from the egg uninfected, at least for B. burgdorferi sensu stricto and B. microti, identification of the larval bloodmeal host through testing of questing nymphs can be used to identify the infecting host and thereby incriminate reservoirs. In this study, every species except vole was identified as a source for B. burgdorferi -infected ticks, but mice and shrews were detected most frequently. Over the 3 transmission seasons, B. burgdorferi was more likely to be acquired from mice in the thicket than mice at the edge (P = 0.05), but the distribution of the remainder of the possible infecting hosts did not differ significantly (Fig. 2). Within specific years, host use differed between the edge and the thicket, and thus the odds of a tick becoming infected after having fed on a particular animal differed. At the edge, ticks that fed on either mice or rabbits were not likely to test positive for B. burgdorferi (Fig. 3), but both were associated with infected ticks in the thicket, although the association for mice was not statistically significant (mouse OR = 2.0, 95% confidence interval of 0.9 to 4.5, and rabbit OR = 6.2, 95% confidence interval of 2 to 19) (Fig. 3). Infected ticks were not likely to have fed on birds, voles, squirrels, and to be determined (TBD) at either site. Ticks with multiple bloodmeal hosts appeared to be more likely to be infected with B. burgdorferi at the edge, but this was not statistically significant (OR = 2.1, 95% confidence interval of 0.9 to 5.0). However, shrews were highly associated with infected ticks (OR = 4.5, 95% confidence interval of 2.2 to 9.1 for edge, and OR = 11.0, 95% confidence interval of 4 to 31 for thicket). Deer were not associated with B. burgdorferi-infected ticks at the edge, and they were significantly less likely to be associated with B. burgdorferi infection in the thicket (P = 0.005).

FIG 2.

Bloodmeal analysis of B. burgdorferi-infected ticks from the edge (A) versus the thicket (B) using the full data set with external contamination removed. The data are presented as a percent of the total number of ticks that tested positive at each site for each host type; all years are combined. The colors of the bars indicate the other species identified in multiple bloodmeals. Asterisks indicate that the percent of ticks with that host type is significantly greater at one site versus the other (Fisher’s exact test, P ≤ 0.05).

FIG 3.

The likelihood that a nymphal tick infected with B. burgdorferi had fed on each host at the edge (A) versus the thicket (B) using the full data with contamination removed. The data are presented in a box plot of the odds ratio with 95% confidence intervals. The dotted line marks an odds ratio of 1, which indicates no association. Confidence intervals that include this line are not statistically significant.

Mice, rabbits, shrews, squirrels, and deer served as sources for B. microti (Fig. 4). No positive ticks were found to have fed on birds or voles. Although there was no significant difference in the overall proportion of ticks from each host from the edge compared to the thicket, the odds of a tick testing positive after having fed on a particular animal were quite different. Mice and shrews were, once again, the main hosts on which infected ticks had fed. However, feeding on mice was not associated at all with infection with B. microti (OR = 0.8, 95% confidence interval of 0.3 to 2.0) at the edge or the thicket (OR = 1.0, 95% confidence interval of 0.4 to 3.1) (Fig. 5). Deer, rabbits, and to be determined were also not significantly associated with infected ticks from either site. However, shrews were very highly associated with infected ticks both from the edge and thicket (OR = 7.9, 95% confidence interval of 3.6 to 17.4 for edge, and OR = 19, 95% confidence interval of 6.1 to 58 for thicket) (Fig. 5).

FIG 4.

Bloodmeal analysis of B. microti-infected ticks from the edge (A) versus the thicket (B) using the full data set with external contamination removed. The data are presented as a percent of the total number of ticks that tested positive at each site for each host type; all years are combined. The colors of the bars indicate the other species identified in multiple bloodmeals.

FIG 5.

The likelihood that a nymphal tick infected with B. microti had fed on each host at the edge (A) versus the thicket (B) using the full data with external contamination removed. The data are presented as a box plot of the odds ratio with 95% confidence intervals. The dotted line marks an odds ratio of 1, which indicates no association. Confidence intervals that include this line are not statistically significant.

DISCUSSION

During the 3 years of this study, we were able to identify the host upon which the majority of the collected nymphal ticks had fed as larvae. Although we were unable to identify the host for 17% of the ticks, only a small percentage of these ticks tested positive for either infection. Therefore, we are confident that we have identified the majority of species that contribute to enzootic transmission of B. burgdorferi and B. microti in our Nantucket study site. Our data illustrate the microfocality of enzootic transmission of tick-borne infections. Distinctly different host contributions to infection were documented in the two contiguous sampling sites, and these contributions changed between years. Similar to what has been previously reported (22), the prevalence of both infections in ticks was greater at the edge than the thicket. There are also differences in host utilization at the two sites. Edge ticks had fed on more diverse species than those in the thicket (Table 3). In particular, rabbits and shrews were more likely to serve as hosts at the edge. The number of ticks that fed on deer was significantly greater in the thicket versus the edge. Deer are known to forage at the edges but bed-down and shelter under the thicket, and there were distinct signs of deer in the thicket at our site, including scat and well-defined deer trails. Ticks that yielded no bloodmeal source (TBD or to be determined) were also more likely to be found in the thicket. Given the known limitations of our bird primers, it may be that these ticks had fed on a species of bird that is more likely to be found in the thicket than at the edge that our assay failed to identify. The bird assay attempts to target diverse species that are not closely related, thus creating a unique challenge for primer design. We have found that these primers have only moderate sensitivity to some target bird species, such as wrens and redstarts, and have greater sensitivity to many others, such as catbird, towhees, and thrushes (39). Avian genomes are smaller and harbor significantly fewer retrotransposons than other vertebrates (40). Alternatively, these unknown ticks could have fed on some other animal that was not included in our assay, Mus musculus or Rattus norvegicus, for example. The latter has been sporadically trapped from our Nantucket study site. Finally, our previous work demonstrated that the success rate of our bloodmeal assay declines as the ticks age (38). It may be that there is a cohort of older ticks in the thicket due to better survival rates there, as the canopy helps to protect the ticks from desiccation. Although there is currently no convenient method for aging of ticks and this hypothesis remains untested, this hypothesis is consistent with the fact that we regularly collected more ticks in the thicket than at the edge.

Our sampling sites could be considered to physically overlap or be an ecotonal continuum, with the edges only a few meters away from the core of the thicket. We could not sample the first meter or so within the edge/thicket interface due to the thickness of the vegetation; we did not attempt to penetrate this brush while dragging the edges of the mown path. We considered the possibility that at this scale of sampling, ticks might move between the edge and thicket. Ticks are known to be passively transported by wind or by phoretic hosts (ticks may infest without attaching, then drop off in a new site) (41, 42). Host-seeking Ixodes ticks actively move, although this has been estimated to be at most an area of 1 m² (41, 43). Unlike metastriate ticks, Ixodes spp. do not move much toward sources of carbon dioxide (44, 45) and thus are less likely to move where hosts are more abundant. Although there may be some passive transport or active movement of ticks between the edge and thicket, our assays identified differences in host contributions to feeding ticks between the two sites, suggesting that when engorged larvae drop from a host, they seek a hiding place very close to that point where they molt and must not move much in a horizontal plane once molted. Accordingly, interventions targeting ticks and their hosts within thickets as opposed to at the edge may have different impacts.

The noted wilderness conservationist Aldo Leopold first used the term “edge effect” (46) to refer to increased game animal diversity and abundance in transitions between habitats. Habitat edges have been described as being biologically more diverse (e.g., in ecology textbooks such as reference 47), although a systematic review of the literature suggests that the edge effect is actually more complex with species abundance of birds and mammals increasing, decreasing, or demonstrating no effect depending on the specific characteristics of the edge (48). In many cases, invasive plant species dominate edges; deer tick abundance can be associated with invasive plants such as barberry, scotch broom, or honeysuckle (49, 50). Goldenrod is a main component of the herb/forb cover at the edges of our site, but the characterization of goldenrod as an invasive species is debated; it is certainly a dominant weed in old fields.

The current paradigm for the force of transmission of deer tick-transmitted infections is that it is an inverse function of increased host diversity, a concept known as zooprophylaxis (9) or dilution effect (51). If edge sites tended to be those of greater activity by a more diverse group of hosts, prevalence of infection should be less than that where reservoir-competent animals, such as P. leucopus, are more abundant. At the Nantucket Field Station, the density of these mice was associated with an increasing gradient of woody stems (52), which distinguishes our thicket site. Although ticks fed on a more diverse array of hosts at our edge site than in the thicket site, the prevalence of infection in ticks was, in fact, significantly greater for both B. burgdorferi and B. microti. At the small scale of our study, deer did not appear to act as a zooprophylactic host, inasmuch as the prevalence of infection did not diminish in 2019 when deer were the primary host for ticks.

We note that the prevalence of infection in ticks was higher at the edge than the thicket. This may be due to the finding that edge ticks more commonly fed on highly competent hosts other than mice, such as rabbits and shrews, than those that fed in the thicket. Surprisingly, despite being thought of as the most competent host, mice were not positively associated with Borrelia infection at the edge (OR = 0.7), and although a number of infected ticks had fed on rabbits, they were also not significantly associated with infection (OR = 0.9). In fact, having fed on either was associated with not being infected with B. microti (OR = 0.8 and 0.3 for mouse and rabbit, respectively). In the thicket, however, these two hosts did have a positive association with infection, though only rabbit was statistically significant (OR = 6.2, 95% confidence interval of 2 to 19). We hypothesize that host diversity, per se, or even the species in particular, is not as important as ticks focusing their bites on a few highly infectious hosts. The 20-80 rule (80% of infections are due to 20% of a host population; also known as overdispersed or aggregated distributions) (53) is relatively ignored in Lyme disease ecology (but see references 54, 55). Despite the odds being quite low that a mouse-fed tick was also infected, mouse bloodmeals were identified in a large proportion of the infected ticks collected in this study, suggesting that there may be a small number of highly infectious individuals in a local deme with the remainder less or not infectious at all (or, more likely, less infested by ticks). We note that most of the infected mouse-fed ticks were identified during our first year of collections when the vast majority of ticks fed on mice. In subsequent years, when fewer mouse-fed ticks were detected, many fewer mice were likely responsible for infecting ticks.

Although it is known that ticks can be groomed off and reattach to a second host (56), the frequency with which this occurs is not known, as there has been no easy way to measure this in nature. Our data indicate that this may occur quite commonly, as we often detected multiple bloodmeal hosts in a single tick. We are not the first to report this phenomenon. Examining bloodmeals from ticks using less-sensitive assays, other investigators also detected multiple bloodmeal hosts in 10 to 23% of the ticks examined (57–60). Although our assay is sensitive enough to be influenced by gross contamination during handling, e.g., by contaminated forceps (38), we do not believe that our results are affected by this due to our strict contamination control measures. The assay is not sensitive enough to detect environmental DNA, but gross contamination from the infestation of a phoretic host (no bloodmeal taken but tick is transported) can be detected (61). To address this possibility, we have removed likely contamination from our data. In 2018, mice were detected in all of the ticks tested, and therefore, were suspected to be a likely contamination source. Therefore, mice were removed from the multiple bloodmeals from that year. Deer appeared to be the majority host in 2019, and the data was treated similarly. There was no majority host for 2020. Because we cannot know whether removing the most likely sources of contamination fixes the issue, we have also analyzed the data with all multiple host ticks removed. The resulting data set is much smaller than the original, thus losing power for identifying associations; these results are included in the figures and tables in the supplemental material. The main findings (the greater diversity of hosts at the edge and the disproportionate contribution of shrews for infected ticks) both remain.

This phenomenon of the detachment of partially fed ticks that then finish feeding on a second host has not been adequately studied and could have consequences for the maintenance of tick-borne pathogens in nature. This would be particularly influential for nymphal tick feeding, if they are frequently detached before they become replete: if an infected tick was attached to the first host for a sufficient length of time for reactivation (62) to occur, the pathogen may be transmitted to the first host as well as the second (63). The second bite would thus be more likely to infect because pathogens are already reactivated if the probability of being groomed off increases as a duration of feeding. Larval deer ticks can acquire infection rapidly (64), and in fact larvae may become infected simply by a quick immersion in B. burgdorferi cultures and subsequently transmit infection when they are allowed to attach to a host (65).

A greater contribution of bloodmeals from mice appears not to be required for spirochetal maintenance at the short time scale of our studies (2 to 4 transmission seasons). Infection prevalence in ticks for both B. burgdorferi and B. microti remained unchanged at both the edge and the thicket in 2019 despite the decrease in the number of ticks that had fed on mice. The prevalence of B. burgdorferi at the edge did decrease in 2020 (from 22% in previous years to 13%), although this difference was not statistically significant and may be simply the result of sample size. While it appears that mice were in fact the major species during the first year of our study, we also found that shrews, rabbits, and deer were important hosts. Over the next 2 years of our study, mice fed only a small percentage of the ticks (<20% for 3 of the 4 collections in 2019 and 2020). Deer appeared to replace mice as the major host for ticks then. Shrews and rabbits were also highly utilized by ticks, especially at the edge. Birds and squirrels appeared to be less important. Because we did not attempt to estimate the densities of any of the mammals in our sampling site, we cannot determine whether host utilization was proportional to within-year species abundance or whether there was preferential feeding. Host substitution has been documented with a related tick, Ixodes persulcatus, which is found to attack rabbits only when its preferred vole host is absent (66). Our data demonstrate that host utilization, as directly measured by bloodmeal remnant identification, is highly dynamic, even on an island with limited fauna, and there can be great variation in between transmission seasons. Such dynamics suggest that short-term interventions targeting mice (e.g., tick tubes, mouse-targeted vaccines) will vary in their efficacy between years.

Despite the expectation that deer are incompetent hosts for both B. burgdorferi and B. microti, we identified a number of infected ticks that had fed on deer. Many of these ticks showed evidence of having also fed on other animals, and it is likely that the infections in those ticks had been acquired from the other more-competent hosts. However, there are a few ticks containing either agent that showed no evidence of a second bloodmeal host. It may be that these ticks had actually fed on a second bloodmeal source that remains unidentified. We were unable to identify the bloodmeal source for 4% to 8% of infected ticks, and thus that bloodmeal source could have been present within what appeared to be ticks solely derived from deer. We must, however, entertain the possibility that ticks feeding on deer were able to acquire infection. Deer are thought to be incompetent hosts for B. burgdorferi (67–69) and their serum is known to lyse spirochetes (70, 71), but there are enigmatic contradictions in the literature. B. burgdorferi-like spirochetes have been reported from deer (72, 73), and naive deer inoculated with B. burgdorferi were able to sustain infection (74). It may be that young or naive deer are able to become infected transiently and serve to infect ticks that are feeding at that time. Transmission could also occur via cofeeding, as this has been demonstrated to occur even on animals that are incompetent reservoirs (75). The possibility of deer being competent reservoir hosts for B. microti seems less probable. Deer have been surveyed extensively for Babesia spp., and although they are certainly infected by other piroplasms, such as Babesia odocoilei or Theileria cervi throughout the eastern United States, B. microti has never been detected in field studies of deer (B. odocoilei is a common Babesia species in Nantucket deer ticks, but our PCR assay is specific for B. microti). Indeed, experiments inoculating deer with B. microti failed to cause infection (76). Cofeeding transmission appears unlikely because B. microti sporozoites delivered during the feeding of an infectious tick would have to have the capacity to be ingested and form secondary kinetes that would survive the molt and invade the salivary glands to form the sporoblast (77). Cofeeding experiments are the only means to rule out the likelihood of sporozoite pluripotency.

Shrews were recognized early on as competent hosts for B. burgdorferi and B. microti, but their importance as a reservoir host was dismissed because the specimens that were caught had few ticks on them (78). Because they are not commonly trapped with the Sherman and Longworth mouse traps that are commonly used for disease ecology studies (but see Adler [52], who trapped on the UMass Field Station in the early 1980s and captured 1 shrew for every 3 to 4 mice), the presence of these hosts is greatly underappreciated. Furthermore, shrews often die in these traps, and the ticks may detach before the animals are examined (personal observation) leading to underestimates of the number of ticks each animal is feeding. Of course, the 20-80 rule would apply to shrews as well. In our study, although the overall number of ticks that fed on shrews was less than either deer or mice, they contributed disproportionately to the enzootic transmission of both B. burgdorferi and B. microti. Shrews were the only species that was significantly associated with infected ticks for both sites. Their importance for the transmission cycle of B. microti is even greater than that for B. burgdorferi (OR of 7.9 at the edge and 19 in the thicket). Although the magnitude of the effect was surprising, this was not completely unanticipated. Looking at B. burgdorferi genotypes that appear to be host specific and extrapolating using a transmission model, Brisson et al. predicted that the contribution of shrews is being overlooked and needed to be reexamined (33). Indeed, molted ticks collected from field-caught shrews yielded large numbers of B. microti-infected ticks (79), and shrews have been found to feed more Ixodes ricinus and more Ixodes trianguliceps than voles (80). Furthermore, shrews are not affected by forest fragmentation, accounting for their common presence in forest patches (81). Blarina brevicauda was previously found to be associated more with habitat characterized by dense grass stems and herbs/forbs (mainly goldenrod) on the Nantucket Field Station (52), which comprises much of the edge there, along with Carolina rose and bayberry. Interestingly, shrews were not the host most commonly utilized for bloodmeals in our study, and this serves, once again, to emphasize the impact small numbers of highly infectious hosts can have on the force of transmission at a site. Unfortunately, we cannot distinguish between individuals and do not know how many different shrews served as hosts for the ticks in this patch. Shrews appear to be capable of sustaining enzootic transmission, perhaps in the absence of mice, and further investigations should focus on this underappreciated host.

MATERIALS AND METHODS

Tick collections.

Host-seeking nymphal ticks were collected from the University of Massachusetts Nantucket Field Station on Nantucket Island, Massachusetts, during June and July of 2018, 2019, and 2020. This 43-ha site has been previously described (38, 82). The nonmarsh area of the site comprises thickets of dense shrub (highbush blueberry, poison ivy, fox grape, greenbrier) with sparse individuals of northern white cedar trees, interspersed with mown fields and walking paths. The edges along these fields and paths comprise mainly grasses, goldenrod, bayberry, and beach rose. The walking paths are well maintained and are 1.5 to 2 m wide; these paths comprise mown grass and bare patches with sandy soil. The understory of a 0.7-ha (Fig. 6) thicket was dragged or swept by physically entering the thicket and stooping or crawling under the canopy. This thicket has been the main tick sampling transect at the UMass Field Station since 1985 (e.g., references 83–85). The thicket has always demonstrated heavy deer use (trails, scat, hoofprints). The edges of the thicket and the outer 0.5 m of the path at the edge were dragged by walking the path that circumscribed the thicket (Fig. 6A and B). The path around the thicket comprised about 300 linear meters. During the drag sampling of the edge, the drag was not used to penetrate the thicket (Fig. 6D) but simply brushed along the edge (Fig. 6C). The thicket and path edge were sampled within the same mornings or afternoons. Ticks from each collection area, thicket, or edge/path were placed into separate vials with hardened plaster on the bottom to provide humidity. All ticks from a single collection were placed together in a vial. Each sampling session (typically >30 min each) was performed during 2 or more consecutive days during June of each year, and the resulting ticks were pooled. In 2018, the ticks were held at 9°C until processing. Subsequent years, the ticks were immediately frozen in their collection vials and kept at −20°C until processing, as previous work has shown an increased failure rate of bloodmeal remnant identification in ticks allowed to age in the lab (38).

FIG 6.

UMass Field Station study site. (A) Thicket area circled in red; blue arrow points to landmark for zoomed in view in panel B. Scale is 100 ft (30.5 m). (B) Zoomed in view of thicket; blue arrow is the landmark from panel A. Scale is 20 ft (6.1 m). (C) Representative vegetation for edge. (D) Representative vegetation for thicket. Modified from Google Maps.

Tick extractions and infection testing.

Ticks were homogenized in phosphate-buffered saline (PBS), and the DNA was extracted by alkaline lysis (HotSHOT) (86). In short, tick homogenates were boiled for 30 min in NaOH-EDTA and then neutralized with Tris-HCl. Ticks were then tested by PCR for evidence of infection with B. burgdorferi and B. microti using a published multiplex assay (87) using SsoFast probes mastermix (Bio-Rad Laboratories, Hercules, CA).

Bloodmeal identification.

Bloodmeal hosts were identified using retrotransposon-based real-time PCR assays that are previously published (38, 39) using SsoFast probes mastermix (Bio-Rad Laboratories, Hercules, CA). Ticks were tested in duplicate. A sample was determined to be positive only if it was repeatable. The assays target mice (Peromyscus), deer (Odocoileus), voles (Arvicolinae), shrews (Soricidae), birds (primarily Passeriformes), squirrels/chipmunks/groundhogs (Sciuridae), and rabbits (Lagomorpha). Although the primers pick up all Sciuridae, red squirrels, chipmunks, and groundhogs are not present on Nantucket, so we are confident that any positives in this study are from gray squirrels. Our assays derive from those used in criminal forensic investigations, which seek to detect trace amounts of highly degraded DNA by targeting amplicons less than 100 bp in length. Sequence data from such small amplicons is comprised primarily from the primers and is not used to confirm the specificity of amplification (88). Instead, the specificity was ensured by testing against a panel of DNA from other animals that could potentially serve as bloodmeal hosts in New England (including house mouse, skunk, raccoon, and rat (see reference 38 for full list and sources)). For most primers, the specificity does not reach the species level and is specific only to genus or family (as specified above). The bird primers were tested against the following 12 most commonly infested birds as reported by Halsey et al. (89): catbird (Dumetella carolinensis), eastern towhee (Pipilo erythrophthalmus), yellowthroat (Geothlypis trichas), ovenbird (Seiurus aurocapilla), Carolina wren (Thryothorus ludovicianus), wood thrush (Hylocichla mustelina), Swainson’s thrush (Catharus ustulatus), Hermit thrush (Catharus guttatus), American robin (Turdus migratorius), veery (Catharus fuscescens), house wren (Troglodytes aedon), and American redstart (Setophaga ruticilla) with varying sensitivity (39).

Because these PCRs are highly sensitive, a great deal of attention was paid to contamination control, including physical separation of tick manipulations, DNA extraction, and PCR amplification. Ticks were manipulated with clean forceps dedicated to this work, both for collections in the field as well as manipulations in the laboratory. Reactions were set up in a clean dead-air hood, and UV light was used for sterilization in between runs. No template controls were included in every step of the process. Primers were used in multiplex reactions to minimize the number of PCRs for each sample. Mouse, vole, and rabbit and then bird, shrew, and squirrel were multiplexed.

Multiple bloodmeals that were identified for individual ticks may be caused by contamination from phoretic hosts; such hosts would not contribute bloodmeals (61). We attempted to reduce confounding of our results by such contamination by assuming that multiple bloodmeals that include the host identified in the overwhelming majority of ticks in the site that year is due to contamination from that majority host; it is less likely that a host less commonly feeding ticks in the site would serve as the phoretic host. Most samples yielding evidence of more than one bloodmeal were thus included in the analysis as derived from a single bloodmeal. Thus, in 2018, mouse, which was detected in 100% of ticks from the edge and 60% of ticks from the thicket, was removed as a contributor from samples for which our assays suggested multiple bloodmeals. In 2019, deer were deleted from multiple bloodmeals; they were detected in 72% of the edge ticks and 57% of the thicket ticks. No host dominated in 2020, so the data was left without adjustment. Any remaining multiple bloodmeals are assumed to be real, and the tick is counted once for each identified host. This data is referred to as the full data with contamination removed. The data was also analyzed with all ticks that tested positive for multiple bloodmeal hosts removed from the analysis completely and is referred to as the truncated data set with all multiples removed (see Fig. S1 to S11 and Table S1 in the supplemental material).

Data analysis.

Odds ratios, prevalence estimates with 95% confidence intervals, and Fisher exact tests were calculated using the StatPages (https://statpages.info/ctab2x2.html). The Shannon index of diversity was calculated using Past 3.0 (90).

Data availability.

Data has been deposited in OSF and is available at https://osf.io/h37b8/?view_only=478f4df4cd8a44619e3c5c1b0033c675.