Introduction
Shifts in the geographic distributions of infectious diseases are currently being observed in transforming ecosystems, highlighting the complex and dynamic interface between landscapes and disease ecology [
1,
2]. For example, tick-borne diseases have been expanding for several years in many parts of the temperate world. In Europe and North America, the distribution and abundance of ixodid ticks and their reservoir hosts has been linked to land use change and climate change [
3‐
5]. At the same time, human behavior risk factors are leading to increased contact with ticks [
6‐
8] and as a result, a larger portion of the human population is being exposed to ticks. This is causing an increase in tick-borne disease incidence rates in many regions of the world [
9‐
12]. While epidemiologists have emphasized the importance of incorporating landscape characteristics into studies of the ecological dynamics of infectious disease emergence, less attention has been directed to the human factors modulating the risk of being exposed to vectors and eventually developing disease [
1].
To effectively manage the risk associated with diseases in different regions and landscapes, it must first be accurately defined, assessed, and the factors that determine it understood [
13]. Risk represents the likelihood that an adverse event will occur, given the consequences it would cause. These consequences depend on the vulnerability of the population of interest to the hazard and are determined by factors such as its level of exposure and coping capacity [
13‐
15]. In the context of tick-borne diseases, the hazard level at a given location or time is typically represented by the number of pathogen-infected ticks present (hereafter referred as the “tick hazard”; [
16]). The density of these ticks in the environment depends on a set of ecological conditions that allow them, their hosts, and the pathogens that circulate between them to complete their life cycles [
17,
18]. Exposure represents the degree to which humans encounter vectors. It is related to land use, accessibility and attractiveness of places where ticks are present [
19]. The consequences of exposure to the hazard are ultimately modulated by a range of social and behavioral factors that determine the coping capacity of the population [
15]. For example, the use of tick repellents decreases the likelihood that an individual will be bitten by a tick, and tick checking decreases the likelihood that an individual will become infected with a pathogen if bitten by a tick. The degree of awareness and adoption of these personal protective measures in populations may vary according to socioeconomic factors and regional endemicity [
20]. Overall, all these elements − hazard, exposure and coping capacity − and the way they interact are likely to vary across landscapes and populations [
15,
21].
Several studies have examined the relationship between the level of forest fragmentation in landscapes and the risk of tick-borne diseases to the human populations present. In North America, this research has been conducted primarily in residential agroforestry landscapes [
16]. Several of these studies have shown that the risk of tick-borne diseases is generally higher in areas where the forest is fragmented than in more homogeneous areas, such as large forest stands or urban areas [
5,
22]. Different mechanisms have been suggested to explain this association. One is that the good adaptation of tick hosts (wildlife species used by ticks for reproduction or serving as reservoirs for pathogen transmission) to the varied habitats present in fragmented landscapes and their concentration in small fragments could enhance enzootic transmission. In addition, the increased presence of areas such as forest-field transition zones (ecotones) are favorable for contacts between human and infected ticks [
3].
In urban areas, the risk is generally concentrated in smaller portions of the territory, such as publicly accessible green spaces and natural conservation parks [
23,
24]. However, we do not know currently whether similar processes linking forest fragmentation to increased tick-borne disease risk also take place in these environments. Indeed, in public nature parks, the causes of forest fragmentation, its general importance, and the way it impacts ticks, wildlife and people distribution and thus the tick-borne diseases risk may be different. The natural or anthropogenic presence of different types of habitats such as herbaceous or shrubby areas, the presence of road and trail networks, or built features (e.g., service buildings, lookouts) are all elements that can lead to forest fragmentation in the context of natural parks [
25]. These elements could influence the distribution of tick hazard, human exposure, or both. For example, trail networks are generally the principal driver of visitors’ spatial distribution across the different areas of a park [
26‐
28]. Trails can also create edge effects, causing changes in the adjacent vegetation, altering abiotic conditions such as light and affecting wildlife and tick presence [
29,
30]. The presence of features like viewpoints, waterbodies or facilities like picnic areas and playgrounds can influence landscape attractiveness for people [
28,
31,
32] and therefore influence the level of the park users’ exposure.
Limited research has been conducted to characterize the risk associated with tick-borne diseases in the context of public parks of North America. In parks of southern Quebec (Canada), Ripoche et al. [
30] found more nymphs in forest habitat adjacent to park trails (measured at points between 20 and 60 m from trails) than directly along the trail edges and higher nymph densities near trails with soil surfaces compared to those with gravel surfaces. Hotspots of high nymph densities were observed in less frequented parts of the parks, while cold spots were located in high-traffic areas such as park entries, trailheads and at park edges, close to residential neighborhoods. Falco and Fish [
33] found lower distances to encountering nymphal or adult
I. scapularis in plots that were randomly sampled throughout parks than in areas of high public use identified by park managers, suggesting that high-use areas were characterized by lower tick densities. In view of these results, the authors proposed that a high human presence could limit the local abundance of tick hosts. Indeed, animals generally respond to human presence in a manner similar to their response to predation, i.e., by avoiding or underutilizing highly disturbed areas [
34]. If fewer hosts are available locally, the probability for ticks to complete their life cycle may be reduced, eventually limiting their local abundance. Through these mechanisms, the risk of tick-borne disease transmission could be influenced by the intensity of human presence in public natural areas. However, in these two studies, space use by people and hosts were not directly measured. Overall, quantitative relationships between the tick hazard and population exposure and fine-scale habitat characteristics in parks have not yet been clearly established. In highly used urban natural areas, improving this baseline knowledge would be particularly relevant to inform local tick-borne diseases risk reduction efforts, for both public health and park managers.
Here, we present a case study of the spatial and temporal variation in risk across a periurban nature park with an emerging risk for Lyme disease (LD) transmission. To do so, we integrate population exposure and tick hazard data. These are respectively represented by each trail’s usage intensity and the density of infected nymphs (DIN) with the LD agent, Borrelia burgdorferi, in the vicinity of the trails. This allows us to create an indicator of the probability of human-tick contact (risk) across the park. We also verify the presence of risk hotspots. From a park risk management perspective, the deployment of risk reduction interventions in these hotspot areas could have a significant positive impact. We explore which features of the park landscape are associated with risk and its two components, hazard and exposure. We hypothesize that forest fragmentation is a determinant of risk distribution across the park. In this park context, the level of forest fragmentation can be represented by indicators such as trail density and the presence of developed areas. First, these elements could generate ecological transition zones (ecotones) between two types of habitats, i.e. zones where tick-host reservoir interactions favour the transmission of pathogens and therefore an increase in the tick hazard. Secondly, a high density of trails and the presence of developed elements could promote the attractiveness and accessibility of the sectors, parameters associated with high visitor traffic and therefore an increase in exposure. Our results will allow for better management of emerging tick-borne diseases in nature parks and contribute to the body of knowledge on the links between fine-scale landscape ecology and the dynamics of tick-borne diseases.
Discussion
While many studies have focused on factors driving tick-borne disease incidence at large geographical scales in North America, very few have investigated simultaneously the ecological and human population factors that may determine risk at finer scales. It is however important to better understand the factors at play at this scale, because it is at this scale that interventions to prevent infections occur. In this study, we demonstrate the applicability and utility of an integrative risk assessment approach to estimate the probability of contacts between visitors and infected ticks, at a local intervention scale in a periurban park environment. The methodology employed allowed for the identification of high-risk areas and periods in park, demonstrating its utility as a planning tool in risk mitigation intervention plans in the context of natural public parks. In addition, we identified biophysical attributes that were associated with the risk levels across a highly visited and newly LD-endemic park. The relationships uncovered between fine scale landscape attributes and spatial variability in risk provide key findings on the ecological determinants of tick-borne disease in recreational parks, which may be a significant source of exposure to tick-borne pathogens for populations in urban areas.
First, we showed that the proportion of forest cover around trails was associated with higher levels of risk and influenced both the visitor density (exposure) and the density of infected nymphs (hazard). These results align with previous research indicating that forested habitats are associated with the population exposure to
B. burgdorferi-infected ticks and high population incidence rates of Lyme disease cases compared to other habitat types (e.g., herbaceous and shrubby environments, agricultural areas, urban areas, or wetlands; [
16,
52]. Trails located in areas where forest cover was dominant were more popular, consistent with the hypothesis that visitors may be more attracted to undisturbed forest areas in parks [
53]. Also, infected nymphs were more abundant in areas where forest cover was dominant. This result is consistent with previous observations associating
I. scapularis density with proportion of forest cover at several geographic scales [
54]. However, in contrast to what was found in other studies performed at regional scales, we did not find a relationship between the DIN and forest fragmentation indicators or high human presence [
55‐
58]. At smaller spatial scales, some studies performed in parks found lower DON or DIN levels in park areas with higher public use [
30,
33]. In contrast, here, we did not observe a correlation between estimated DIN and trail use levels. We also found no relationship between DIN and indicators of the territory's accessibility such as distance to entrances, or indicators of forest habitat fragmentation such as edge and trail density or the size of the forest patches. It has been suggested that human disturbance of habitat may affect host presence and tick survival, and thus, that the lowest tick densities would be found in forested areas with the greatest current or past human presence [
30,
33]. The opposite thesis has also been put forward, that the main hosts of blacklegged ticks (white-footed mice, white-tailed deer) adapt well to disturbed habitats and thus may become dominant there at the expense of other wildlife species, which are more sensitive to habitat disturbances [
22]. Thus, since ticks would have more opportunity to encounter reproductive and reservoir hosts, their survival and reproduction, as well as the circulation of tick-borne pathogens, would be favored, resulting in a higher level of tick hazard [
22]. It appears that in our context, neither of these hypotheses apply. Overall, there is relatively low mammal biodiversity across the park, with mice and deer present and abundant in the majority of areas [
59]. The lack of relationship between DIN and fragmentation indicators therefore suggests that the main hosts, mice and deer, are not affected by fragmentation at this fine scale, and are present at sufficient abundances in a range of habitats throughout the park [
59]. However, we did observe a positive relationship between the proportion of forest cover and the DIN. This suggests that in this context, it is the presence of a habitat favorable to the survival of ticks when they are off-host, i.e. a forest floor where the abiotic conditions (temperature, relative humidity, presence of refuges under the leaf litter) necessary for their survival are present, that is the main determinant of their distribution here [
59].
Based on these results, we are suggesting approaches to manage the risk associated with ticks in natural parks. We recommend that actions aimed at decreasing the likelihood of human-tick contact be taken in areas of parks where forest cover dominates. For example, more emphasis should be placed on encouraging the adoption of safe behaviors by users, particularly in forested areas of parks. These best practices include staying on trails, using tick repellents, wearing long clothing, and practicing tick checks after a forest activity [
20,
60]. These practices could be reminded to visitors directly in the parks, through signage in high-risk forest areas, to reinforce their adoption levels. Second, we recommend that trail edges in high-risk areas be landscaped so that the likelihood of human-tick contact is restricted. Trail maintenance that discourages contact includes regular trimming of vegetation along trails and removal of dead leaves from the ground [
61] and installation of wood chips on the ground along trail borders [
62]. Finally, reducing the probability of human-ticks contact can also be achieved by reducing the tick hazard in high-risk habitats. For this, possible interventions include the selective use of acaricides applied to vegetation in high-risk areas and host-targeted interventions (e.g., treatments with acaricides [
63,
64]). However, all these interventions are resource-intensive, which currently limits their deployment over large areas such as the territory of natural parks [
65]. Therefore, we propose here to deploy them first in high-risk areas to optimize the cost–benefit of deployment. Our cluster analyses have shown the presence of risk hotspots. By strategically prioritizing interventions to these hotspots, we could act on 41 to 43% (depending on the year) of the risk in the entire park, while deploying resources to only 11% of the territory. Such hotspots, when present in a park, are therefore places where high impact potential is possible if interventions are deployed. Slight variations in the location of high-risk areas were however present in this study, consistent with previous findings of heterogeneous patterns of tick densities at small spatial scales [
30,
59,
66]. Therefore, periodic reassessment of the location of high tick density areas should be included in park risk management plans. This would ensure that interventions are always deployed in areas where their impact is expected to be the highest.
Second, we showed that proximity to certain facilities was associated with elevated risk. Specifically, we found certain park features (refuges, viewpoints and entrances) associated with increased levels of exposure and risk. This result is consistent with other studies that have found these types of elements to be associated with the level of attractiveness or accessibility of public natural areas [
26,
28,
53]. This finding offers another opportunity for risk management, that of targeting population exposure patterns. Indeed, parks could be designed so that a mismatch between the location of areas of high population exposure and high tick hazard is induced. This could be accomplished by modifying the attractiveness and accessibility of areas, which are important drivers of visitor distribution across territories [
19,
67]. As part of this approach, parks could develop trails, refuges, and lookouts in areas with low tick densities, so that people will tend to use these areas more. They could also limit public access to areas with high tick densities, especially during high-risk periods.
The main risk period in this study corresponded with the peak in nymph abundance [
68,
69]. A second, lower peak in risk was present in early fall, when entomological risk is lower, but when the park receives large number of visitors. We did not, however consider a possible increase in entomological hazard associated with the presence of adult ticks in the fall and thus, the second peak could be underestimated by our analysis. In addition, tick densities were lower in 2018 than in 2017. While it is theoretically possible that tick removal by our sampling efforts could have influenced these results, it appears more likely that this variation is caused by weather fluctuations during these years [
70], likely due to high summer temperatures and reduced rainfall in 2018. Indeed, our results are aligned with trends observed in active surveillance which showed higher generalized tick densities in 2017 than in 2018 across the province [
71,
72].
Here, we conceptualized different levels of risk based on the presence and abundance of pathogen-infected vectors and the human population at risk, in the same space–time. As illustrated above, the approach developed here could be used as a decision support tool for risk management in the context of public parks. However, we cannot conclude on the link between the observed patterns and the actual incidence of LD in the local population. Several other factors could influence the actual patterns of disease acquisition by the local population, such as individual behaviors (i.e., what people are doing in the natural space and for how long), levels of knowledge and perception of risk in relation to the disease [
7,
20], use of tick bite prevention methods or availability of medical management following a tick bite [
73]. In particular, we could not consider how user behavior might influence their level of exposure to ticks in the environment. For example, we could assume that the actual exposure of people using the wide trails is less than for the narrow trails. In the former case, users may come into contact with the vegetation surrounding the trail or leaf litter at the edge of the trail less frequently than when using narrow trails. In the former case, users may come into contact with the vegetation surrounding the trail or leaf litter at the edge of the trail less frequently than when using narrow trails. Since trail edges can be favorable micro-habitats for ticks [
30], different exposure to them could impact users' actual exposure to ticks. In addition, bicycle use was only allowed on the wide trails in the park. If it also turned out that cyclists were less exposed to tick bites than pedestrians, then our estimate of the level of risk on the wide trails could be underestimated. On the other hand, it is also possible that our analysis underestimates risk in some areas of the park. For example, if certain attributes attracted people to go off trails, or to pick things off the ground, these areas could be a source of greater exposure to ticks. Such observations were made in a French park, where areas conducive to picking plants and mushrooms represented the highest risk of exposure to
Ixodes ricinus ticks for users [
74]. Further studies would be needed to assess the effect of users' behaviors on their individual risk of exposure to ticks. Another limitation of this study is the precision of the pathogen prevalence estimates used in the calculation of the tick hazard indices. Ideally, these estimates should be calculated at the site level, but due to limited sample sizes and testing capacity, site-level estimates were too uncertain. Instead, we opted to use park-level estimates, providing higher confidence. While this approach is supported by previous studies showing that the main driver of local DIN is generally DON [
75,
76], larger sample sizes in future studies would increase confidence in local risk assessments. Finally, the risk levels obtained here at the scale of the park would also need to be validated with accurate disease acquisition data. However, the level of detail that would be required on such a fine scale is difficult to obtain in the context of vector-borne diseases. Indeed, the locations of acquisition of infection by vector-borne diseases are often unknown or confidential. Modeling studies could then be used to simulate the dynamics of encounters between the populations described here and the subsequent steps that could lead to the development of the disease in bitten individuals, for example. The proposed strategy of prioritizing intervention in high-risk areas should also be tested in different epidemiological scenarios to demonstrate its effectiveness.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.