Government policies are driving landscape-scale changes in land use to provide ecosystem services; but there may be unconsidered cascading effects. A major land use change targeted for regulating climate and biodiversity is peatland restoration. This will cause changes in vegetation and, potentially, keystone species such as large herbivores, which are the main hosts to Ixodes ricinus (L.) ticks, the most important vector of disease-causing pathogens in Europe.
This study tested the impact of restoring peatlands from conifer forestry on Ixodes ricinus abundance and explored the likely mechanisms. Large-scale surveys of Ixodes ricinus, vertebrate herbivores and vegetation were conducted in adjacent areas of forest, bog and areas felled 5–13 years previously.
Questing tick abundance was greatest in forest and almost absent from blanket bog, with intermediate numbers in felled areas, where ticks were more abundant in young than in old felled areas.
The likely mechanisms for these variations in tick abundance were deer habitat preferences (bog was the least preferred habitat) and ground vegetation height or canopy cover, which are generally associated with alternative tick hosts and micro-climates that aid tick questing and survival.
Synthesis and applications. Felling conifer forest to restore peatlands could produce a dramatic decline in tick abundance throughout the restoration process, with implications for disease risk. Therefore, a further ecosystem service of peatlands in addition to climate, biodiversity and water regulation is regulating pests and disease. Deer management and procedures that speed up the restoration process are likely to enhance the effect during the intermediate stages.
Ixodes ricinus (L.) ticks are the most important vectors of disease-causing pathogens in Europe and are currently changing in abundance and distribution (reviewed by Medlock et al. 2013). They can carry pathogens, such as Borrelia burgdorferi sensu lato, the agent of Lyme borreliosis, the tickborne encephalitis complex of viruses, Babesia species, Anaplasma species and Rickettsia species. Within any given climate type, I. ricinus abundance is influenced greatly by habitat (due to tick host habitat preferences and the micro-climate produced by different habitats; e.g. MacLeod 1935; Clark 1995; Gray 1998; Dobson, Taylor & Randolph 2011), such that we might expect land use changes to alter significantly I. ricinus abundance and distribution (Medlock et al. 2013). One current major landscape-scale land use change is peatland restoration that is being carried out for the purposes of climate change mitigation and biodiversity enhancement.
Peatlands are disproportionately important carbon stores: although they cover only 3% of land surface, they are estimated to contain a third of all carbon held in soils globally (Gorham 1991) and most intact peatlands sequester carbon, with potential for climate change mitigation (Clymo, Turunen & Tolonen 1998; Baird et al. 2009; Worrall et al. 2011). However, damaged and degraded peatlands can do the opposite, becoming net sources of carbon to the atmosphere (Waddington, Warner & Kennedy 2002; Worrall et al. 2011). Vast areas of peatlands globally have been destroyed or degraded through various means, such as pollution, overgrazing, burning, draining, extraction for fuel or compost, afforestation and, more recently, wind farms (Stewart & Lance 1983; Tallis 1998), and less than 20% of the UK's peatlands remain undamaged (Bain et al. 2011).
In the UK, the most common peatland habitat is blanket bog, a rain-fed system characterized by extensive coverage of Sphagnum moss species, along with higher plants such as deer grass Trichophorum cespitosum (L.) Hartm., cotton grass Eriophorum species, heather Calluna vulgaris (L.) Hull and cross-leaved heath Erica tetralix (L.). In the UK, blanket bog is one of the most extensive semi-natural habitats, with approximately 1·8 million hectares, of which >1 million hectares are in Scotland (over 20% of Scotland's land surface; Joint Nature Conservation Committee 2011). Blanket bog is crucially important for conserving biodiversity and is a protected habitat under the EC Habitats Directive Annex 1 and included in the UK Biodiversity Action Plan as a Priority Habitat (Littlewood et al. 2010). Because peatlands provide the ecosystem services of regulating climate, greenhouse gases, carbon and wild species diversity and supporting biodiversity (UK National Ecosystem Assessment 2011; Whitfield et al. 2011), the UK Biodiversity Action Plan has a target for the restoration of peatlands, which, for Scotland, is around 600 000 hectares (see Holden, Chapman & Labadz 2004; O'Brien, Labadz & Butcher 2007). Of all the modes of peatland damage, afforestation changes peatlands the most, requiring drainage, ploughing and tree planting, thereby creating a completely new ecosystem. How will restoring afforested peatlands affect I. ricinus abundance and tickborne disease risk?
Ixodes ricinus have three active stages, larvae, nymphs and adults, that each require one blood meal from a host that lasts a few days, after which they drop off and moult to the next stage and emerge up to 15 months later (depending on the temperature, MacLeod 1934; Randolph et al. 2002). Larvae and nymphs feed on a range of host types from small mammals and birds, mountain hares Lepus timidus (L.) and roe deer Capreolus capreolus (L.) (e.g. Humair, Rais & Gern 1999; Kiffner et al. 2010); adults feed primarily on large hosts such as deer and livestock that are often termed ‘tick reproduction hosts’ because the adult females that feed on them then produce thousands of eggs to continue the next generation of ticks (Gray 1998). In terms of tick hosts in afforested and intact peatlands, densely planted commercial conifer forests, which are typically planted on peatlands, often lack understorey or ground vegetation and have low biodiversity (Moore, Allen & Hunter 1999; Palik & Engstrom 1999), while undamaged blanket bog is of importance to breeding wading birds (Littlewood et al. 2010). Both forests and blanket bog may be used by roe or red deer Cervus elephus (L.). The process of changing forest back into blanket bog includes felling trees and filling the plough furrows with the small trees/brash wood (the furrows were originally created by deep ploughing to drain the bog prior to planting, with the saplings planted on the raised plough-throw between the furrows). The dead tree material helps to infill part of the furrow void and main collecting drains are also blocked to raise water-tables closer to the ground surface so that the original blanket bog vegetation will recolonize the furrow bottom and grow upwards, thereby raising the water-table simultaneously (Price, Heathwaite & Baird 2003). Of particular interest is how tick abundance may change with the different ages of restoration felling (i.e. the process of changing freshly felled forest into a felling–bog transition habitat), with a view to determining how long it takes for tick abundance to be affected. The key objectives of this study were to test the impact on I. ricinus tick abundance of changing forest back into blanket bog, via different ages of restoration fellings, to explore how long any change in abundance might take and to elucidate the likely mechanisms for the effects found.
Materials and methods
This study used cross-sectional surveys of multiple adjacent areas of unfelled forest, restoration fellings (of varying ages) and undamaged blanket bog.
The study was conducted at Forsinard Flows, northern Scotland (58o35′70″ N, 3o89′690″ W), a large (21 000 Ha) nature reserve managed by the Royal Society for the Protection of Birds (RSPB), Scotland. The reserve comprises a mosaic of large areas of open blanket bog, forest and restoration felling. One key aspect of the reserve management has involved acquisition of neighbouring conifer plantations on deep peat and systematically felling these large blocks of forest (planted in the 1980s) for the purposes of blanket bog restoration. This has resulted in large areas (around 2400 ha) where the forest has been felled in one of the UK's largest landscape-scale restoration projects. There is an excellent chronosequence with restoration fellings varying in age (at the time of surveying) from 5 to 13 years of age (from forest felled between 2006 and 1998). Forsinard is therefore a unique ‘natural experiment’, providing the opportunity to study the impact of landscape-scale changes in land use from forest, through different ages of restoration felling, to the eventual target habitat of actively growing blanket bog. Sufficient time has not yet elapsed for the regeneration of intact blanket bog, but here I use undamaged blanket bog to compare with adjacent forest and restoration fellings, since this is the aspirational target habitat.
Eight areas within the Forsinard RSPB reserve were surveyed and, for most, all three habitat types (forest, restoration felling and blanket bog) were surveyed to provide a direct comparison between habitats. However, because not all habitats were available within close proximity in all areas, forest was surveyed at six areas, blanket bog at seven areas and restoration felling at all eight areas. For felled sites, as many different ages were surveyed as possible within any one area (see Table 1 for details).
Table 1. Summary of habitats and restoration ages (year forest was felled) for each of the eight survey areas within the Forsinard site
Year forest was felled
Bog, felling, forest
Bog, felling, forest
Bog, felling, forest
2003, 2004, 2005
Bog, felling, forest
Bog, felling, forest
1998, 2004, 2005
Surveys of ticks, host dung and ground vegetation were conducted three times to encompass the main tick questing season for northern Scotland: in June, July and August 2011. Questing ticks were counted by dragging a 1 × 1 m2 piece of white blanket material over the ground for 10 m, then turning the blanket over and meticulously counting all ticks (Gray & Lohan 1982; Gilbert 2010). At the start and end of each 10-m drag, a circular area 1 m in diameter was carefully inspected for the presence or absence of host dung (at this site, these were primarily red deer, with some roe deer, mountain hares and red grouse Lagopus lagopus scoticus Lath, also present). The mean of the two dung counts was used for analysis. While dung counts of this nature cannot provide an estimate of host density, they can be useful in providing an idea of the usage by each host of each habitat, or an ‘index of relative abundance’ between areas (Gilbert 2010). For analysis, dung counts from both roe and red deer were combined since it was not always easy to distinguish between them. Too few dung from mountain hares and grouse were counted for inclusion in statistical models. While small mammals and birds can be important hosts of I. ricinus larvae (Gray 1998; Dobson, Taylor & Randolph 2011; James et al. 2011), no resources were available to estimate the abundances of these hosts. Ground vegetation influences hosts, creates a micro-climate that can influence tick questing and survival, and may affect the efficiency of the blanket drag technique (Dobson, Taylor & Randolph 2011) and must be taken into account in statistical models (Ruiz-Fons & Gilbert 2010). Therefore, at the start, middle and end of each 10-m drag, ground vegetation height was measured using a sward stick to the nearest 5 cm. The average value over the three height measurements for each drag was used for analysis. Temperature and humidity are key determinants of tick host-seeking behaviour (e.g. MacLeod 1935; Randolph & Storey 1999; Perret et al. 2000; Randolph et al. 2002). Therefore, the temperature and relative humidity were measured at the time of surveying using a temperature/humidity recorder placed on the vegetation (or on the ground if there was little vegetation; range: 0–50 cm) in an attempt to record what questing ticks are experiencing at the time of sampling. It was important to include these in statistical models to help account for conditions that might discourage ticks from actively questing and therefore being counted on blankets, for example, if it is too cold or dry for tick activity.
For each of the three visits (June, July and August), 15 blanket drags were conducted in each habitat type in each of the eight areas surveyed. Tick questing can vary greatly between days, presumably partly due to weather conditions (Perret et al. 2000), so to minimize this effect each habitat within any one area was surveyed on the same day.
For statistical analysis, generalized linear mixed models were conducted using the GLIMMIX procedure in sas version 9.3. A Poisson-log-normal model was specified to account for the data distribution as is usual for ecological count data. First, the effect of habitat (forest, restoration felling and blanket bog) on tick abundance was tested. This was to gain a broad understanding of how tick numbers are likely to change from extant conifer forest, through restoration felling, and ultimately to the target habitat of blanket bog. Secondly, using data from restoration felling areas only, the effect of the age of fellings on tick abundance was tested, to gain an understanding of how tick abundance may vary as restoration felling changes over time into a more blanket bog-like transition habitat.
Analyses were conducted only on counts of nymphs rather than on larvae or adult ticks because fewer larvae and adults were counted and this caused an extremely zero-inflated data distribution, making statistical analysis and data interpretation difficult. In addition, nymphs are the stage considered to pose the greatest risk of tick bites to humans (Robertson, Gray & Stewart 2000).
To investigate differences in the relative abundance of questing I. ricinus nymphs between the three habitats (forest, restoration felling and blanket bog), the explanatory variables that initially entered into each model included the following fixed effects: habitat, deer dung index, month of sampling, vegetation height, temperature and relative humidity recorded at the time of each drag as well as an interaction term between deer and habitat.
To investigate differences in the relative abundance of questing I. ricinus nymphs between the different ages of restoration felling, the same model was used, except that restoration felling age was entered as a fixed effect (continuous variable) and habitat was not entered in the model (because only data from restoration felling areas were used for this model).
All models included area as a random effect, since several blanket drags, habitats and ages of restoration felling were surveyed within each area.
For each analysis, a backwards stepwise procedure was conducted, whereby non-significant (P >0·05) explanatory variables were sequentially removed from the model, starting with interaction terms. Because non-significant terms were eliminated from the models, test statistics are presented for only those fixed effects that remained in the final model.
Finally, potential mechanisms for the effects found were explored with further generalized linear mixed models of deer abundance and general linear mixed ground vegetation height and relative humidity. Deer can be key drivers of tick abundance (Wilson et al. 1990; Stafford, Denicola & Kilpatrick 2003; Rand et al. 2004), including for I. ricinus in Scotland (Ruiz-Fons & Gilbert 2010), and ground vegetation can also affect tick abundance, questing and survival by creating a mild and humid micro-climate and providing habitat for hosts. Therefore, these further models explored the effect of habitat type and restoration felling age on the index of abundance of deer, ground vegetation height and relative humidity. Deer dung counts, similar to counts of questing nymphs, displayed a Poisson distribution so the GLIMMIX procedure in sas was used and a Poisson-log-normal model specified. Ground vegetation height and relative humidity approximated to a normal distribution so general linear mixed models were used with the MIXED procedure in sas. As for the previous models, area was included as a random effect.
Questing Nymph Abundance in Forest, Restoration Felling and Bog
The number of questing nymphs counted per 10 × 1 m blanket drag was 2·64 (±0·73; mean ± SE) in forest, 1·18 (±0·38) in restoration felling and 0·12 (±0·10) on blanket bog, which represents a reduction in questing nymphs of 55% and 95% from forest to restoration felling to bog, respectively (Fig. 1). These differences in questing nymphs with habitat were highly significant (forest had more nymphs than both restoration felling (t1,1005 = 12·53, P <0·0001) and bog (t1,1005 = 13·44, P <0·0001) and restoration felling had more nymphs than bog (t1,1005 = 8·65, P <0·0001; see Fig. 1)). In addition, there were more nymphs counted where there were more deer (Fig. 2), higher ground vegetation and when the relative humidity was higher (Table 2). There was also an effect of month of sampling (July had more questing nymphs than both June (t1,1005 = 3·21, P =0·0039) and August (t1,1005 = 9·64, P <0·0001) and June had more nymphs than August (t1,1005 = 6·63, P <0·0001; Table 2)). The relationship between questing nymph abundance and deer abundance index differed between habitats (the deer*habitat interaction was significant; Table 2), such that there was a significant positive effect of deer on nymph abundance in restoration felling (F1,559 = 9·64, P =0·0020), but not bog (F1,218 = 1·30, P =0·2558) or forest (F1,218 = 0·17, P =0·6800). Temperature was eliminated from the model due to non-significance.
Table 2. Final models of factors influencing questing nymph abundance index to test for (a) habitat differences including data from all habitats (top half of the table) and (b) the effect of felling age using data from felled areas only (bottom half of the table). Model coefficients, degrees of freedom (DF), test statistic (F) and significance (P) are shown. For categorical variables (habitat and month), coefficient values are in relation to a baseline value of zero (bog and felling values are in relation to forest; July and August are in relation to June). Values are not shown for explanatory variables eliminated from the model during the backwards stepwise procedure
(a) All habitats
(b) Felling only
Questing Nymph Abundance with Restoration Felling Age
There was a highly significant effect of restoration felling age, with more questing nymphs counted in areas of recently felled forest and fewest nymphs in older fellings (Fig. 3; Table 2). Other effects on nymphs were as for the previous analysis, that is, there were more nymphs counted where there were more deer (as indicated by dung counts), higher ground vegetation and when the relative humidity was higher and there was a highly significant effect of the month of sampling (July had more nymphs than August (t1,558 = 9·05, P <0·0001) and June had more nymphs than August (t1,558 = 8·12, P <0·0001), although there was no significant difference between June and July (t1,558 = 1·08, P =0·5273; Table 2)). Temperature was not significant and eliminated from the model.
How Similar to Blanket Bog is the Oldest Restoration Felling?
The two oldest areas of restoration felling (both 13 years old) had, on average, 0·13 and 0·15 nymphs per blanket drag in contrast to their respective adjacent areas of forest that had 1·80 and 2·11 nymphs per blanket drag, representing reductions in questing nymph abundance of 92·6% in both areas. The respective adjacent areas of undamaged blanket bog had 0·00 and 0·08 nymphs per blanket drag, representing reductions in questing nymph abundance of 100% and 95·8%, which are only slightly greater reductions than for the adjacent 13-year-old restoration felling areas.
Patterns of Deer Abundance, Ground Vegetation Height and Relative Humidity with Habitat
There was a highly significant association between habitat and the index of abundance of deer (F2,1012 = 27·26, P <0·0001; Fig. 1), with far fewer dung counted on blanket bog than in either forest (t1,1010 = 6·39, P <0·0001) or restoration felling (t1,1010 = 7·29, P <0·0001). There was no significant difference in deer dung counts between forest and restoration felling (t1,1010 = 0·25, P =0·8032).
There was a highly significant association between restoration felling age and the index of abundance of deer, with far more deer dung counted in areas of newer restoration felling (recently felled forest) than in areas of older restoration felling (F1,329 = 16·20, P <0·0001; Fig. 3).
There were significant differences in ground vegetation height between habitats (F2,1005 = 204·7, P <0·0001; Fig. 1). There was far less ground vegetation in forest (that usually had just bare ground covered in pine needles) than in restoration felling, which contained grasses (t1,1005 = 20·10, P <0·0001), or blanket bog that was moss-dominated (t1,1005 = 13·54, P <0·0001), and bog had significantly shorter vegetation than restoration felling (t1,1005 = 3·68, P =0·0007). Within the restoration felling habitat, the ground vegetation height decreased significantly with restoration felling age, that is, after accounting for random effects, older restoration felling generally had lower vegetation than younger felled areas (F1,556 = 7·35, P =0·0069) although the effect size was small (Fig. 3).
There was a marginally significant association of habitat with relative humidity recorded at the time of tick surveying (F2,1005 = 3·09, P =0·0459), whereby bog was slightly more humid than forest (t1,1005 = 2·35, P =0·0492), but not restoration felling (t1,1005 = 0·80, P =0·7053), and there was no evidence for a difference in humidity between restoration felling and forest (t1,1005 = 2·01, P =0·1092). There was no significant association of relative humidity at the time of surveying with restoration felling age (F1,556 = 2·90, P =0·0889).
The aims of this study were to test how questing I. ricinus tick abundance changes with peatland restoration from commercial coniferous forest via different ages of restoration felling, using a replicated chronosequence of plots at the RSPB Forsinard Flows Nature Reserve. Potential mechanisms (deer abundance, ground vegetation and relative humidity) for the patterns found were also explored. Restoration felling areas were associated with a 55% reduction in nymphs and undamaged blanket bog with a 95% reduction in nymphs compared to forest. This strongly suggests that restoring blanket bog from conifer forest will progressively reduce I. ricinus abundance until virtual eradication once blanket bog is fully restored, and has implications for disease risk.
However, because the oldest restoration felling areas surveyed were only 13 years old and are still undergoing restoration management, the time period needed for restoration felling areas to become blanket bog is not known. It has been estimated that the re-establishment of important bog species can be achieved in 5 years from the start of a peatland restoration scheme, a high water-table re-established in around 10 years and an ecosystem that accumulates peat in around 30 years, although this is for cutover (mined) peatlands that have been spread with diaspores, mulched and fertilized (for rapid re-colonization of Sphagnum mosses and vascular plants) as well as drain-filled (Rochefort et al. 2003). In general, not enough time has yet elapsed in peatland restoration projects to determine how long it takes to restore the original ecosystem structure, function and biodiversity, and the time needed (or whether restoration will occur at all) will depend on the type and level of damage and the methods of restoration (Gorham & Rochefort 2003). There is no information on the success or time required for restoring blanket bog from forest by placing felled trees or brash into furrows. However, in terms of tick abundance, the two oldest areas of restoration felling (both 13 years old) both had 92·6% fewer questing nymphs than adjacent forest compared with 100% and 95·8% reductions in the respective adjacent areas of undamaged blanket bog. This suggests that, while 13 years is not enough time for complete blanket bog restoration, it may be almost enough time, at least in some areas, to reduce tick abundance to levels found in blanket bog.
What are the potential mechanisms for the reduction in ticks as forest is transformed into blanket bog, via different ages of restoration felling? Deer are the primary reproduction hosts for ticks in many parts of the Northern Hemisphere (e.g. Gray 1998), including for I. ricinus in Scotland (Gilbert et al. 2012). It is therefore not surprising that deer (roe and red) abundance was significantly associated with questing I. ricinus nymph abundance in this study, and this concurs with many other studies both for roe and red deer and I. ricinus (Kiffner et al. 2010; Ruiz-Fons & Gilbert 2010; Gilbert et al. 2012) and for other deer and tick species (Wilson et al. 1990; Stafford, Denicola & Kilpatrick 2003; Rand et al. 2004). Accordingly, there was very little deer dung found in blanket bog compared with restoration felling or forest, which can explain the lack of ticks on blanket bog. However, deer abundance cannot adequately explain why there were more ticks in forest than in restoration felling areas. Other environmental factors associated with questing nymph abundance in this study were ground vegetation height (higher vegetation was associated with more ticks) and relative humidity at the time of surveying (higher humidity was associated with more questing ticks counted). Ground vegetation structure and type is known to influence the abundance of alternative tick hosts, such as birds and small mammals (e.g. Smit et al. 2001), and create different micro-climates that affect tick questing and survival (MacLeod 1935). Ground vegetation height decreased with restoration felling age and was low in blanket bog, similar to questing nymph abundance, suggesting that vegetation as well as deer abundance may be a mechanism explaining nymph abundance. However, while nymphs were most abundant in forest, ground vegetation was shortest (often bare ground) in forest. There are two possible reasons why there were more questing nymphs in forest than in restoration felling, despite lower ground vegetation in forest and no difference in deer abundance between forest and restoration felling. First, and most likely, is that the canopy cover of the densely planted coniferous forest is likely to have produced a micro-climate favourable for tick questing and survival; this is considered a key reason why I. ricinus are often more abundant in forest than in open habitats (Walker et al. 2001; Lindström & Jaenson 2003). Secondly, the lack of ground vegetation in forests may cause the blanket drag survey technique to record a higher proportion of the questing tick population than in tall (or dense) vegetation (Dobson, Taylor & Randolph 2011) although entering vegetation height into the statistical models should have helped to allow for this.
What are the implications for tickborne disease risk? The risk of being bitten by an infected tick depends on tick abundance and the prevalence of the pathogen in the tick population, which in turn is influenced by the relative numbers of competent transmission hosts. Clearly, if restoring blanket bog from forest is associated with a >95% reduction in I. ricinus abundance, disease risk is also likely to be dramatically diminished. In terms of B. burgdorferi s.l., the agent of Lyme borreliosis, the key transmission hosts (small mammals and birds) are also likely to be less numerous on bogs than in forests or restoration felling areas, thereby even further reducing disease risk. However, further research would be needed to determine densities of transmission hosts and pathogen prevalence in questing ticks in order to quantify the change in risk. Incidence of infection also depends on the number of people frequenting an area. In general, forests attract more recreational visitors than blanket bogs, and most cases of Lyme borreliosis in Scotland are contracted in woodland or forest habitats (James 2010), further suggesting a greatly reduced disease risk as afforested peatland is restored. At this study site, Forsinard Flows RSPB reserve, more visitors are eventually anticipated as a result of afforested peatland restoration, due to the attractive landscape and biodiversity. However, this is unlikely to result in higher infection rates; such is the predicted virtual disappearance of ticks on blanket bog compared to forest and fellings.
In conclusion, this study provides compelling evidence that successful restoration of blanket bog from forestry will result in the virtual eradication of I. ricinus in a time-scale of around 13 years, and the likely mechanisms are deer habitat preferences and factors associated with ground vegetation and forest canopy cover (such as micro-climate and small tick hosts). Further management such as furrow blocking and brash crushing that speed up the restoration process is likely to increase the rate of tick decline. Peatland restoration is being conducted primarily to improve the ecosystem services of regulating climate and greenhouse gases and supporting biodiversity. This study suggests further ecosystem services of regulating pests and/or diseases through dramatic declines in I. ricinus abundance and improving cultural health benefits through reduced tickborne disease risk.
I am indebted to David Fergusson for conducting fieldwork, Norrie Russell for support and permission to work on the RSPB reserve at Forsinard Flows and Rebekka Artz for initial advice on the sites. Norrie Russell and David Johnson gave valuable comments on the manuscript. This research was supported by the Scottish Government Rural and Environment Science and Analytical Services Division.