INTRODUCTION
In recent decades, Lyme borreliosis has become a subject of international concern because of the increasing number of human cases diagnosed each year (World Health Organization, 2004; Bacon et al. Reference Bacon, Kugeler and Mead2008). The disease is caused by spirochetes belonging to the Borrelia burgdorferi sensu lato (s.l.) complex, which is maintained in an enzootic cycle involving mainly ticks of the Ixodes ricinus species complex (Acari: Ixodidae) and numerous vertebrates (Piesman and Gern, Reference Piesman and Gern2004). Several bird and small mammal species, rodents in particular, serve as hosts for the larval and nymphal stages and are important reservoirs for Borrelia spirochetes (Kurtenbach et al. Reference Kurtenbach, Peacey, Rijpkema, Hoodless, Nuttall and Randolph1998; Humair et al. Reference Humair, Rais and Gern1999; Comstedt et al. Reference Comstedt, Bergström, Olsen, Garpmo, Marjavaara, Mejlon, Barbour and Bunikis2006). Ticks may acquire the spirochetes by feeding on infected hosts, maintain the infection to the subsequent life stages through trans-stadial transmission, and transmit the infection to other hosts or humans during the next bloodmeal.
The risk of human infection is typically associated with forested areas, as forests provide ticks with access to a broad range of vertebrate hosts and favourable environmental conditions for tick survival (Gray, Reference Gray1998). It is commonly assumed that the observed increase in Lyme borreliosis incidence is due to an actual rise in the number of infections and not only to an enhanced surveillance and awareness of the disease. This has been attributed to various factors, particularly to human encroachment into forested areas and habitat modifications, resulting in a closer human contact with ticks and an increase in the abundance and range of tick populations. For instance, the marked increase of deer populations throughout Europe and North America during the latter half of the twentieth century (Gill, Reference Gill1990; Cederlund et al. Reference Cederlund, Bergqvist, Kjellander, Gill, Gaillard, Boisaubert, Ballon, Duncan, Andersen, Duncan and Linnell1998), largely caused by changes in land use and land cover (e.g., reforestation), has been ascribed a major role in the emergence and spread of Lyme borreliosis (Spielman, Reference Spielman1994; Sood et al. Reference Sood, O'Connell, Weber and Sood2011). Although incompetent reservoirs for B. burgdorferi, deer are the preferred hosts of adult Ixodes ticks and play an important role in their reproductive success. Moreover, both tick abundance and infection prevalence in ticks are favoured by forest fragmentation (Allan et al. Reference Allan, Keesing and Ostfeld2003; Brownstein et al. Reference Brownstein, Skelly, Holford and Fish2005; Halos et al. Reference Halos, Bord, Cotté, Gasqui, Abrial, Barnouin, Boulouis, Vayssier-Taussat and Vourc'h2010; Tack et al. Reference Tack, Madder, Baeten, Vanhellemont, Gruwez and Verheyen2012), since deer and rodents benefit from the presence of forest edge habitat and abundant ecotonal vegetation (Tufto et al. Reference Tufto, Andersen and Linnell1996; Saïd and Servanty, Reference Saïd and Servanty2005; Boyard et al. Reference Boyard, Vourc'h and Barnouin2008). In Sweden, I. ricinus has become more abundant and has expanded its range during the last 3 decades, which is probably caused by changes in climate (increased duration of the vegetation period and milder winters) and increased abundances of deer (Jaenson et al. Reference Jaenson, Jaenson, Eisen, Petersson and Lindgren2012).
In many European countries, the conversion of monospecific coniferous forests into mixed, structure-rich forests dominated by native broadleaved species has become a major objective of sustainable, multipurpose forest management, with the aim of optimizing the production of various goods and ecosystem services (Olsthoorn et al. Reference Olsthoorn, Bartelink, Gardiner, Pretzsch, Hekhuis and Franc1999; Spiecker et al. Reference Spiecker, Hansen, Klimo, Skovsgaard, Sterba and von Teuffel2004). However, deciduous forests, especially those harbouring significant numbers of cervids, are generally considered to be ideal habitats for I. ricinus, which is the most common tick species associated with Lyme borreliosis in Europe (Gray, Reference Gray1998). By altering the forest composition and structure, these forest management activities involve a large-scale land-use change that might influence the suitability of forests for ticks and, consequently, might influence the epidemiology of tick-borne diseases (i.e., forest conversion creating an ecosystem dysfunction). Yet, there have been relatively few studies addressing the variation in tick abundance between forest types. While it has recently been quantitatively shown that the abundance of I. ricinus ticks is higher in oak stands compared to pine stands and increases with increasing shrub cover (Tack et al. Reference Tack, Madder, Baeten, Vanhellemont, Gruwez and Verheyen2012), little is known on the longer-term temporal variation and on the effects of forest composition and structure on the resulting Borrelia infection rate. Here, we selected a total of 25 pine (Pinus sp.) and oak (Quercus sp.) stands, both with and without a substantial shrub layer, and sampled I. ricinus tick populations between April and October in 3 successive years in northern Belgium to describe the spatiotemporal variation in the abundance of larvae, nymphs, and adults and to relate this variation to forest composition and structure. Additionally, habitat use by cervids was determined by counting the number of deer beds. Borrelia burgdorferi s.l. spirochete infections in tick nymphs, the life stage predominantly responsible for pathogen transmission, were determined by polymerase chain reactions to assess the potential impact of forest conversion on the infection prevalence.
MATERIALS AND METHODS
Study area
The study was conducted at 2 forest sites in the Campine ecoregion in northern Belgium. Forest site A (51°17′ N, 5°12′ E) was located near the border with the Netherlands in the municipality Postel and forest site B (51°2′ N, 4°58′ E) was located approximately 30 km to the south in the municipalities Herselt and Tessenderlo. The climate is sub-atlantic: the mean annual precipitation amounts to 799 mm and is evenly distributed throughout the year, with mean monthly precipitation ranging from 53 mm in March to 79 mm in July. The mean annual temperature is 9·0°C, with minimum and maximum mean monthly temperatures of 1·4°C and 16·7°C in January and July, respectively (Royal Meteorological Institute of Belgium, URL http://www.kmi.be/, accessed November 18, 2011). The region's characteristic forests are pine plantations—mainly consisting of Scots pine (Pinus sylvestris) and, to a lesser extent, Corsican pine (P. nigra subsp. laricio)—on nutrient-poor and acidic sandy soils. The pine stands are interspersed with deciduous stands of pedunculate oak (Quercus robur), red oak (Q. rubra), common beech (Fagus sylvatica), silver birch (Betula pendula), and downy birch (B. pubescens) (Waterinckx and Roelandt, Reference Waterinckx and Roelandt2001). Most forests were established in the nineteenth and first half of the twentieth century on former heathlands, which once formed an important component of the traditional agricultural system and covered most of the landscape. The then prevailing microclimatic conditions (temperature and moisture) were most likely limiting for tick survival, which is strongly supported by recent studies carried out in heathlands (Estrada-Peña, Reference Estrada-Peña2001; Lindström and Jaenson, Reference Lindström and Jaenson2003; Wielinga et al. Reference Wielinga, Gaasenbeek, Fonville, de Boer, de Vries, Dimmers, Akkerhuis Op Jagers, Schouls, Borgsteede and van der Giessen2006). However, the large-scale afforestation and the subsequent rise in deer populations probably made this region suitable for tick population establishment and survival. Nowadays, the Campine region is known as a hotspot area in Belgium for Lyme borreliosis (Linard et al. Reference Linard, Lamarque, Heyman, Ducoffre, Luyasu, Tersago, Vanwambeke and Lambin2007). Local vertebrate hosts of nymphal and female ticks are large and medium-sized mammals such as roe deer (Capreolus capreolus), red fox (Vulpes vulpes), European hare (Lepus europaeus), European hedgehog (Erinaceus europaeus), least weasel (Mustela nivalis), European pole cat (Mustela putorius), and red squirrel (Sciurus vulgaris). Very common small mammalian hosts for larvae include pygmy shrew (Sorex minutus), common shrew (Sorex araneus), wood mouse (Apodemus sylvaticus), bank vole (Myodes glareolus), and field vole (Microtus agrestis) (Verkem et al. Reference Verkem, De Maeseneer, Vandendriessche, Verbeylen and Yskout2003; Tack et al. unpublished data).
Forest stand selection
At each forest site, 6 pine stands and 6 oak stands were selected on poor, sandy soils with half of the stands having little or no shrub layer (<15% shrub layer cover in the 1–7 m height class) and the other half having a well-developed shrub layer (>50% cover). An additional oak stand was selected with low shrub cover at forest site B. In summary, ticks were sampled in 25 forest stands (12 in forest site A and 13 in forest site B) and in 4 distinct forest stand types: pine stands and oak stands, both with and without a substantial shrub layer. The relative contribution of Pinus sp. (P. sylvaticus or P. nigra) or Quercus sp. (mainly Q. robur) to the total estimated canopy cover of the tree layer (>7 m) was greater than or equal to 80% in each pine and oak stand, respectively. In each forest stand, the percentage cover of the shrub layer (1–7 m) and herb layer (<1 m) was estimated visually. Shrub cover estimates were very comparable between pine and oak stands at both forest sites. The structure-rich oak stands had an average shrub cover of 66·7% at site A and 70·0% at site B, and the pine stands had an average shrub cover of 70·0% at site A and 58·3% at site B. The shrub layer mainly consisted of alder buckthorn (Frangula alnus), black cherry (Prunus serotina), and rowan (Sorbus aucuparia) in the pine stands and alder buckthorn, pedunculate oak, and sycamore (Acer pseudoplatanus) in the oak stands. The herbaceous layer was dominated either by wavy hair-grass (Deschampsia flexuosa), purple moor-grass (Molinia caerulea), broad buckler-fern (Dryopteris dilatata), or bilberry (Vaccinium myrtillus), providing a comparable blanket contact when drag sampling for ticks (see below). Forest stands with a dense bracken (Pteridium aquilinum) understory were avoided because this vegetation can seriously impede tick sampling (Tack et al. Reference Tack, Madder, De Frenne, Vanhellemont, Gruwez and Verheyen2011). Because of the height and rough vegetation surface of bracken, ticks are easily brushed off the blanket, causing tick abundance to be underestimated. However, the sampled vegetation types are representative for the Campine region so we do not expect our sampling procedure to greatly affect the results.
Sampling strategy
Tick sampling was carried out between April and October in 2008, 2009, and 2010 for a total of 11 occasions at site A and 12 occasions at site B (12 stands×11 occasions +13 stands×12 occasions=288). Sampling consisted of dragging a white flannel blanket (1×1 m2) over the herbaceous vegetation and litter. In each forest stand and at each sampling occasion, we performed 6×1 min blanket drags (each extending a distance of ca. 25 m) at random and recorded the air temperature and relative humidity 3 times at a height of 1·25 m above the soil surface, using a portable digital temperature and relative air humidity meter (DM509, Eijkelkamp Agrisearch Equipment, Giesbeek, The Netherlands). Sampling was always performed on dry (no rain) and non-windy days (<2 Bft) during day time (between 10:00 am and 05:00 pm) when the vegetation was dry. To avoid time of day and changing meteorological conditions as a source of bias, the 4 forest stand types were sampled in random order on each sampling day. After each transect, larvae, nymphs, and adults were removed from the blanket using forceps and stored in vials containing 70% ethanol for later identification and counting. The ticks were counted and identified morphologically with a stereo-microscope using the identification keys of Hillyard (Reference Hillyard1996).
Additionally, the number of fecal pellet groups and beds of roe deer were counted at each sampling occasion, along the same transects used for tick sampling. Pellet-group counting is a widely used method for assessing habitat use by deer. In our study, however, the number of pellet groups counted was too small (only 21 pellet groups in total) for proper analysis. Instead, we have used the number of deer beds in each forest stand type to examine differences in habitat selection for bedding sites (Smith et al. Reference Smith, Oveson and Pritchett1986; Bíró et al. Reference Bíró, Szemethy, Katona, Heltai and Petö2006). Deer beds were easily detectable in the sandy soil of the study area and were distinguished as oval depressions in the soil or as flattened areas of vegetation, often accompanied by other signs of roe deer (e.g., hoof prints, hair).
Identification of Borrelia infections
Twenty pooled samples per forest site per year (20 samples×2 sites×3 years=120), with each sample consisting of 5 nymphs, were used for further molecular analyses for the presence of B. burgdorferi s.l. spirochetes. We did not identify the Borrelia genospecies. Instead, only screening up to species level was performed to get an idea of the overall infection prevalence. For each forest site and each year, 10 samples consisted of nymphs collected in pine stands with low shrub cover while the other 10 samples were collected in oak stands with high shrub cover. Potential differences in infection prevalence are most likely to occur between these two contrasting forest stand types. DNA was extracted using the method of Boom et al. (Reference Boom, Sol, Salimans, Jansen, Wertheim-van Dillen and van der Noordaa1990). This method is based on the lysing and nuclease-inactivating properties of proteinase K, together with the nucleic acid-binding properties of silica particles. A standard PCR amplification was performed in 25 μl reaction mixtures containing 5 μl of the extracted DNA, 1·65 mM MgCl2, 0·2 mM of all 4 dNTPs, 10 pM of 2 primers (BorrSLospAF/BorrSLospAR) (Demaerschalck et al. Reference Demaerschalck, Ben Messaoud, De Kesel, Hoyois, Lobet, Hoet, Bigaignon, Bollen and Godfroid1995), 1 UTaq polymerase enzyme (Promega), and 1 μl of Yellow SubTM (GENEO Bioproducts, Hamburg, Germany). After a hot start of 10 s at 84°C, an initiation of 4 min at 92 °C was performed and followed by a 40-cycle denaturation-hybridization-elongation step (30 s at 92°C, 45 s at 58°C, and 60 s at 72°C). The PCR ended with an extension step of 10 min at 72°C. Then 5 μl of each reaction mixture were mixed with 2 μl of loading buffer and loaded onto 2% agarose gels (Sigma) to be examined for the presence of DNA fragments. A 1·5 kb DNA ladder (MBI Fermentas, Lithuania) was loaded on every gel. The samples were run for 20 min at 100 V, stained in ethidium bromide for 30 min, washed under running tap water, and photographed under UV illumination.
Statistical analysis
Questing tick abundance, expressed as the number of ticks collected per 100 m2, was first log10(n+1) transformed to approach normality, which was verified using the Kolmogorov-Smirnov test. Subsequently, log-transformed tick abundances were modelled with linear mixed models using the lmer-function of the lme4-library (Bates et al. Reference Bates, Maechler and Bolker2011) in R 2.13.0 (R Development Core Team, 2011). Data for each life stage (larva, nymph, and adult) were analysed separately. Models included tree species (pine vs oak), shrub cover (in%), herb cover (in%), year, and all their two-way interactions as fixed effects and forest stand (nested within forest site (A or B)) and sampling occasion as non-nested random-effect terms. To analyse the effects of tree species, shrub cover, year, and all their two-way interactions on the presence of roe deer (scored as 1 or 0 depending on whether deer beds were (1) or were not (0) encountered in the forest stand while dragging), we applied a generalised linear mixed model (GLMM) with similar random-effects structure as above, but with a binomial error distribution and logit link function. Analysis of nymphal infection with B. burgdorferi s.l. (pooled samples of nymphs infected (1) or not (0)) were also performed with a GLMM with binomial error distribution and logit link function. This model included forest type (pine stands with low shrub cover vs oak stands with high shrub cover), year, and their interaction term as fixed effects and forest stand (nested within forest site) as random-effect term. We always compared all possible models (i.e., build by each combination of the fixed-effects terms) using Akaike's Information Criterion, adjusted for sample size (AICC) (Hurvich and Tsai, Reference Hurvich and Tsai1989). The ΔAICC of a model was then calculated as the difference in AICC value for that model and the model with the lowest AICC value (best fit to the data). Models with ΔAICC ⩽4 were considered equivalent (Bolker, Reference Bolker2008). To determine the relative importance of the explanatory variables, we used the sum of Akaike weights of the set of all top models (ΔAICC ⩽4) in which the variable appeared (Burnham and Anderson, Reference Burnham and Anderson2002). The Akaike weight reflects the weight of evidence in support of a particular model relative to the entire model set, and varies from 0 (no support) to 1 (complete support). Finally, the parameter values of the model with the lowest AICC value were estimated with restricted maximum likelihood estimation.
RESULTS
In total, 110 770 I. ricinus ticks were collected, of which 89 017 were larvae, 18 685 were nymphs, and 3068 were adults (1634 males and 1434 females). During tick collection, the air temperature ranged from 7·1°C to 31·7°C and the relative humidity ranged from 28·1% to 92·6%. The mean±standard error of the number of ticks collected per 100 m2 was 206·1±20·8 larvae (range 0 to 4263), 43·3±2·1 nymphs (range 1 to 215), and 7·1±0·4 adults (range 0 to 44). On each sampling occasion, all 3 life stages were active and ticks were found questing in all 25 forest stands studied. In May 2009, a very high number of larvae was collected along a single transect in one of the oak stands with high shrub cover, which resulted in a peak in larval activity in May (Fig. 1a). This high variance in larval abundance was not unexpected and reflects the limited dispersal capability of larvae after emergence from the egg mass, consisting of up to 2000 eggs (Jongejan, Reference Jongejan2001). By considering this single transect as an outlier, questing larvae showed a summer peak (August) each year. Nymphs were active throughout the study period without displaying a clear peak (Fig. 1b). Adult tick abundance peaked in spring (April–May) each year and steadily declined in summer (Fig. 1c). Our data were not suited to study seasonal variation in tick abundance, but our results are in line with those of Gassner et al. (Reference Gassner, van Vliet, Burgers, Jacobs, Verbaarschot, Hovius, Mulders, Verhulst, van Overbeek and Takken2011), who examined the temporal dynamics of I. ricinus in a neighbouring country, The Netherlands.
Fig. 1. Mean number of Ixodes ricinus larvae, nymphs, and adults (a–c) and mean number of deer beds (d) in pine and oak stands between May and October in 3 successive years. The results from the 2 forest sites were pooled. Error bars denote the standard error of the mean. Note the difference in values on the y-axis.
For both larvae and adults, the best model explaining the variation in tick abundance included tree species and shrub cover as explanatory variables (Table 1). For adults, a second closely competing model also included a tree species by year interaction term. The best model for nymphs included tree species, shrub cover, and year, whereas the second best model included only tree species and year (Table 1). Herb cover did not appear in any of the top models. Tree species, on the other hand, was present in all top models of each life stage and was therefore the variable with the highest relative importance in explaining tick abundance (Table 2). The temporal fluctuations in tick abundance were very similar in oak and pine stands, but the mean abundance was consistently higher in the oak stands (Fig. 1a–c; Table 3). Larvae, nymphs, and adults were on average 3·3, 1·6, and 1·5 times more abundant in the oak stands. Shrub cover was also a variable of high relative importance (Table 2) and had a positive effect on tick abundance (Table 3). On each sampling occasion, the mean number of ticks collected was higher in forest stands with high shrub cover compared to stands with low shrub cover. Overall, the number of larvae, nymphs, and adults was 2·1, 1·5, and 1·8 times higher in forest stands with high shrub cover (>50% cover) (Fig. 2a–c). Hence, mean tick abundance was lowest in pine stands with low shrub cover (43·3±8·2 larvae, 20·7±2·0 nymphs, and 3·8±0·5 adults per 100 m2) and highest in oak stands with high shrub cover (418·9±72·2 larvae, 61·6±4·9 nymphs, and 11·1±1·0 adults per 100 m2) (Fig. 2a–c).
Fig. 2. The effects of tree species and shrub layer cover on the number of Ixodes ricinus larvae, nymphs, and adults (a–c) and on the number of deer beds (d) in 3 successive years. Shrub cover estimates were grouped into 2 classes: low (<15%) and high (>50%) cover. The results from the 2 forest sites were pooled. Error bars denote the standard error of the mean. Note the difference in values on the y-axis.
Table 1. Model selection statistics for the analyses of effects of tree species (T), shrub layer cover (S), and year (Y) on the abundance of Ixodes ricinus larvae, nymphs, and adults and on the presence of deer beds
Table 2. Relative importance of each explanatory variable, calculated across all top models (ΔAICC ⩽4, see Table 1) in which the variable appeared
Table 3. Parameter estimates (P.E.) of the best model (see Table 1) for the abundance of Ixodes ricinus larvae, nymphs, and adults and for the presence of deer beds
A very similar pattern was observed regarding the number of deer beds we encountered during tick sampling (Figs 1d and 2d). The best model explaining the presence of deer beds included tree species, shrub cover, and year (Table 1), with the first two being the variables with the highest relative importance (Table 2). The probability of encountering deer beds was significantly higher in oak stands (n=288, P=0·006) and in forest stands with high shrub cover (n=288, P=0·015) (Table 3). The mean number of deer beds was 1·6 times higher in forest stands with high shrub cover and twice as high in oak stands, which resulted in 4 times as many deer beds in oak stands with high shrub cover compared to pine stands with low shrub cover.
Borrelia-positive nymphs were found each year at both forest sites. The average infection rate with B. burgdorferi s.l. was 8·3% (95% confidence interval: 4·8–13·2%) in 2008, 11·3% (7·0–16·9%) in 2009, and 6·2% (3·4–10·7%) in 2010. A similar infection rate was observed at both forest sites in the first 2 years of our study, but the infection rate at site A (1·0%) was considerably lower compared to site B in 2010 (12·9%). No significant difference in infection rate was observed between the homogeneous pine stands and the structure-rich oak stands (n=120, P=0·850). The average infection rate was 8·3% (5·4–12·2%) in the pine stands and 8·7% (5·7–12·8%) in the oak stands.
DISCUSSION
Our results show that tree species composition and vertical structure are important variables in explaining tick abundance in forests. The abundance of all 3 life stages was higher in oak stands compared to pine stands, and increased with increasing shrub cover. Interestingly, this pattern was observed at both forest sites and on almost every sampling occasion. So, although some annual and seasonal fluctuation in tick numbers occurred, the mean tick abundance was always lowest in the homogeneous pine stands and almost always highest in the structure-rich oak stands. On average, the abundance of larvae, nymphs, and adults was 9·7, 3·0 and 2·9 times higher in the oak stands with high shrub cover than in the pine stands with no or little shrub cover, while intermediate abundances were recorded in the 2 remaining forest stand types. The observed differences in tick abundance between the forest stand types must not necessarily depend directly on differences in tree species composition or structure, but may rather be caused by differences in activity of host animals. Our observations from deer bed counts indicate that roe deer were more often present in oak stands and in stands with high shrub cover, most likely because of the availability of high-quality forage and shelter. The importance of deer in maintaining tick populations has been stressed in several European studies (Gray et al. Reference Gray, Kahl, Janetzki and Stein1992; Pichon et al. Reference Pichon, Mousson, Figureau, Rodhain and Perez-Eid1999; Ruiz-Fons and Gilbert, Reference Ruiz-Fons and Gilbert2010). Being the most common large mammals in the study area, roe deer are almost certainly the most important hosts for adult ticks and, therefore, their habitat use largely determines the location where engorged female ticks drop off and lay eggs. The immature stages (larvae and nymphs) also feed on large mammals such as roe deer, but they generally feed on small to medium-sized mammals and birds. Rodents, such as bank vole and wood mouse, have been identified by several authors as key hosts for larval ticks (Tälleklint and Jaenson, Reference Tälleklint and Jaenson1997; Humair et al. Reference Humair, Rais and Gern1999; Estrada-Peña et al. Reference Estrada-Peña, Osácar, Pichon and Gray2005). These rodent species, together with other mammal species such as foxes and hedgehogs, are common in the study area and provide immature ticks the opportunity to successfully obtain a bloodmeal and develop into the next life stage, which explains the relatively high nymphal and adult abundances in our study.
Besides being important hosts for immature ticks, small mammals and birds are also important reservoir hosts for Borrelia spirochetes. Borrelia afzelii has been associated with mice, voles, and red squirrels, B. burgdorferi sensu stricto with red squirrels, and B. garinii and B. valaisiana mainly with birds (Humair and Gern, Reference Humair and Gern1998; Kurtenbach et al. Reference Kurtenbach, Peacey, Rijpkema, Hoodless, Nuttall and Randolph1998; Humair et al. Reference Humair, Rais and Gern1999; Hanincová et al. Reference Hanincová, Schäfer, Etti, Sewell, Taragelová, Ziak, Labuda and Kurtenbach2003). The different genospecies tend to cause distinct clinical manifestations affecting different systems (van Dam et al. Reference van Dam, Kuiper, Vos, Widjojokusumo, de Jongh, Spanjaard, Ramselaar, Kramer and Dankert1993) and, thus, the vertebrate host composition will determine not only the density of Borrelia-infected ticks but also the relative risk of different clinical forms of Lyme borreliosis. We did not identify the Borrelia genospecies, which could be considered a shortcoming of our study. However, B. afzelii and B. garinii, both known to be pathogenic to humans, are the two most common Borrelia species in Belgium, the Netherlands, and northern France (Rauter and Hartung, Reference Rauter and Hartung2005), suggesting that most larvae feed on small rodents and birds in this region. A study carried out in the Netherlands (Gassner et al. Reference Gassner, Verbaarschot, Smallegange, Spitzen, Van Wieren and Takken2008) showed a significantly higher nymphal abundance and Borrelia infection rate in oak plots than in pine plots, which was ascribed to differences in rodent densities. In our study, however, the nymphal infection rate with Borrelia varied substantially for the different forest sites and years, but no significant effect was found for forest type. Yet, as the absolute number of ticks was considerably higher in oak stands and stands with an abundant shrub layer, the chance of getting bitten by ticks and acquiring infection is in fact influenced by forest type.
The results of this study have important implications for forest management, as management activities can alter the composition and structure of forests, which could have a profound impact on the epidemiology of tick-borne diseases such as Lyme borreliosis. In response to environmental concerns and changing societal needs, one of the main goals of the forest management policy in many parts of Europe is the conversion of (often coniferous) plantations to semi-natural forest types. To achieve this, large areas of homogeneous coniferous stands are being converted into mixed, structure-rich deciduous stands with oak as one of the main constituents. Our results indicate that this forest type can support higher tick population levels than monospecific plantations.
However, whereas tick abundance was highly affected by tree species and shrub cover, the overall Borrelia infection rates in ticks were similar in the 2 contrasting forest types. On the other hand, it is important to note that monospecific pine stands cover most of the area in both forest sites, while oak stands, especially those with an abundant shrub layer, are relatively scarce. Large-scale forest conversion programmes could change the composition and abundance of wildlife communities to the extent that the relative proportion of reservoir-competent and incompetent hosts changes, thereby influencing not only tick abundance but the infection prevalence in ticks as well. In the past decade, increasing attention has been paid to the role of biodiversity in mediating infection levels and disease, termed the dilution effect (Ostfeld and LoGiudice, Reference Ostfeld and LoGiudice2003). The current study underlines the importance of considering spatial heterogeneity in forest habitat quality when studying tick populations and supports vegetation management as a tool to control tick populations. Relatively simple interventions such as mowing the vegetation and clearing brush along forest trails have been shown to be effective in reducing the local abundance of ticks (Wilson, Reference Wilson1986; Schulze et al. Reference Schulze, Jordan and Hung1995). However, further studies will be required in order to fully understand the effects of forest conversion on Lyme borreliosis risk.
ACKNOWLEDGEMENTS
The authors owe special thanks to Miguel Lyssens-Danneboom, Kris Ceunen, and Filip Ceunen for assistance with fieldwork and to Margot Vanhellemont and Robert Gruwez for critical reading of the manuscript. We are also grateful to Natuurpunt vzw and the Flemish Nature and Forest Agency (ANB) for the permission to work in the forests.
FINANCIAL SUPPORT
This work was supported by IWT-Flanders, the Institute for the Promotion of Innovation through Science and Technology in Flanders.
