To graze or not to graze? Sheep, voles, forestry and nature conservation in the British uplands

Authors


Darren Evans, A Rocha International, 3 Hooper Street, Cambridge CB1 2NZ, UK (e-mail darren.evans@arocha.org).

Summary

  • 1The British uplands are internationally important for their unique plant and bird communities. They have considerable economic, nature conservation, landscape, aesthetic and tourism-related value and as a result are under a variety of different pressures, most notably from sheep farming, shooting interests, commercial forestry and conservation interests.
  • 2In order to optimize biodiversity in the uplands, the challenge is to find how to balance the different land-use pressures. One key upland species that is potentially affected by livestock grazing, and is of considerable interest to both foresters and nature conservationists, is the field vole Microtus agrestis. Relaxation of livestock grazing can result in an increase in vole numbers. This in turn could have both positive and negative implications for biodiversity as (i) field voles are a major source of prey for other species and (ii) they are a cause of damage to newly planted trees and potentially damaging in areas of native woodland regeneration.
  • 3A replicated, randomized block experiment, consisting of six replicates of four livestock grazing treatments, was established in 2003 (with baseline data collected in 2002). This enabled us to examine the effects of livestock grazing on field vole abundance.
  • 4We have demonstrated experimentally for the first time that livestock grazing pressure affects the abundance of field voles (as measured by a vole sign index) in the uplands. In the first year of the experiment, immediate treatment effects were detectable, with a lower abundance of voles in the conventionally grazed treatment compared with those in the ungrazed treatment, and with intermediate vole abundances in the lightly grazed treatments. The significant treatment effects became more apparent in 2004, with a higher abundance of voles in the extensively grazed mixed treatment (i.e. sheep and cattle) than in the extensively grazed treatment that contained only sheep.
  • 5Synthesis and applications. In order to maximize biodiversity in the uplands, our results suggest that low intensity livestock grazing, used in a novel ways, could be a useful management tool to reduce vole abundance (and hence subsequent tree damage) compared with excluding livestock from young plantations completely. Furthermore, vole abundance would be higher at low grazing intensity compared with conventional stocking rates, thus still providing food for raptors and other vole-eating vertebrates. Indeed, low-intensity sheep and mixed livestock grazing might improve prey availability as a direct consequence of increased heterogeneity in vegetation structure. Our results suggest that it may be possible to maintain the open character of moorland habitats, and benefit key upland species generally, by reducing sheep grazing pressure and introducing low-intensity mixed livestock grazing throughout the uplands.

Introduction

The British uplands (in particular the heather moorlands) are of high nature conservation value. There are 19 constituent plant communities, of which five are virtually confined to Britain and 13 are listed under the EC Directive 92/43/EEC on the Conservation of Natural Habitats and of Wild Fauna and Flora. British moorland also supports 40 species of birds, eight of which are listed under Annex 1 of the EC Directive 79/409/EEC on the Conservation of Wild Birds. Approximately 20% of upland heather moorland present in England and Wales in the mid-1940s has changed under afforestation, agricultural reclamation, high grazing pressure and bracken Pteridium aquilinum invasion. Of that remaining, 70% is estimated to be at risk of change (Thompson et al. 1995). As a result, the regeneration and conservation of heather moorland has received much attention in recent years (Thompson & Brown 1992; White & Wadsworth 1994; Thompson et al. 1995).

Grazing by large herbivores and controlled burning (e.g. to create suitable habitat for game birds) are principal factors maintaining the open character of upland habitats (Fuller 1996). Subsequent long-term changes in heath and moorland vegetation have been closely linked with changes in grazing pressure (Stevenson & Thompson 1993). Since the 1970s, there has been a large increase in overall sheep numbers in several European Union (EU) countries (Beaufoy, Baldock & Clark 1994). In Britain, between 1950 and 1990 sheep numbers more than doubled, from 19·7 million to 41·2 million (Fuller & Gough 1999). Despite much local variation, this increase has coincided with a shift from systems of mixed herbivores, cattle, goats, sheep and horses, towards ones dominated by sheep within many upland regions (Sydes & Miller 1988). This increase in sheep is implicated in dramatic changes in vegetation and bird abundance (Woods & Cadbury 1987; Ratcliffe 1990; Thompson et al. 1995; Fuller & Gough 1999; Evans et al. 2005). Hence there have been calls to reduce sheep numbers to halt the decline of moorland biodiversity (Thompson et al. 1995). It is not only sheep that have increased. In Scotland, the number of red deer Cervus elaphus has also steadily increased since the 1960s and it has been suggested that the population doubled between 1965 and 1995 (Harris et al. 1995). The pattern of grazing has, however, never been static, both in historic and recent periods grazing has been an agent of profound ecological change in the British uplands (Fuller 1996). Similarly, in North America grazing by domestic livestock has fundamentally altered native biological communities and a vigorous debate now centres on this land management issue (Fleischner 1994; Brown & McDonald 1995; Fondell & Ball 2004).

Commercial forestry is another important land use in the British uplands, and the area of woodland is generally increasing in all parts of the UK (Anonymous 2003). Livestock are frequently excluded from new plantations to prevent tree damage. Furthermore, to limit damage by large, wild herbivores such as red deer and roe deer Capreolus capreolus, fencing up to 1·8 m in height is often used to exclude them from a plantation, especially when deer density is high (Hodge & Pepper 1998). The fertilization and planting of large tracts of land with exotic conifers, and the resulting disturbance of existing plant and animal communities, has historically been a source of conflict between foresters and nature conservationists (Avery & Leslie 1990; Petty & Avery 1990). Additionally, fencing to exclude livestock and deer from woodlands often causes the death of birds of conservation concern, such as capercaillie Tetrao urogallus, that can collide with them in flight (Moss 2001). Throughout Europe, however, forests are currently undergoing changes in form and intensity of management, with a concomitant desire to examine and promote the role of large herbivores in the maintenance of traditional woodland pasture systems as well as high forest (Kirby, Mitchell & Hester 1994; Hester et al. 2000).

In order to optimize biodiversity in the uplands, the challenge is to find how to balance the different objectives of sheep farmers, foresters, sporting estates and nature conservationists. One key upland species that is of considerable interest to both foresters and nature conservationists is the field vole Microtus agrestis (L.), a widely distributed rodent found predominantly in rough grassland throughout the Palearctic (Gipps & Alibhai 1991). It has been suggested that the relaxation of livestock grazing can result in an increase in vole numbers (Hill, Evans & Bell 1992; Hope et al. 1996). This in turn could have both positive and negative implications for biodiversity as (i) field voles are important prey for vertebrate predators and (ii) they can be a major cause of damage to newly planted trees (especially broadleaves).

Field vole population cycles frequently occur in northern parts of its range, such as Scotland and northern Scandinavia (Gipps & Alibhai 1991). The breeding density and breeding success of raptors such as hen harriers Circus cyaneus, kestrels Falco tinnunculus, short-eared owls Asio flammeus and barn owls Tyto alba have been shown to be closely linked with vole abundance (Village 1987, 1990; Taylor 1994; Petty 1999; Redpath, Thirgood & Clarke 2002; Petty & Thomas 2003). However, it is not known to what extent grazing reduces small mammal populations and thus the primary source of food for many avian predators. At least in one species, the hen harrier, increased sheep grazing has been implicated as a cause of population decline (Meek et al. 1998; Amar, Redpath & Thirgood 2003). The interaction between livestock grazing and vole abundance, and how this influences the breeding success of predatory birds, is still unclear. Crucially, it is not known to what extent livestock grazing affects not only vole abundance but the availability of voles to raptors as a result of changes to vegetation structure.

Field voles are of concern to commercial foresters during the early stages of planting as they strip bark off the lower stem and roots of young trees, which are often killed as a result. In one of few experiments looking at how the use of herbicides might reduce vole damage to trees generally, Davies & Pepper (1993) found that 86% of unweeded sycamore Acer pseudoplatanus were damaged by voles. Damage to trees is most likely to occur when vole numbers are high and food is scarce, but serious damage can occur at any time of year (Davies & Pepper 1993). New plantations are fertilized to encourage tree growth and virtually all upland sites receive one application of ground mineral phosphate at the start of the rotation (Avery & Leslie 1990). The combination of both fencing (to exclude stock) and fertilizers is thought to result in significant vole population growth, often described as vole ‘booms’ (Avery & Leslie 1990). Vole ‘plagues’ (c. 2–3000 ha−1) have also been reported in commercial forests (Lockie 1955; Charles 1956). Vole damage is, however, of particular concern in the creation of new native woodland or when encouraging natural regeneration of broadleaves. Bank voles Clethrionomys glareolus have been shown to damage small-leafed lime Tilia cordata saplings and seedlings of oak (Quercus spp.) and beech Fagus sylvatica (Pigott 1985). According to Harmer (1995) damage to naturally regenerated broadleaves by voles may be underestimated. Indeed, there have been few studies that have looked at the effects of vole damage to deciduous plantations and areas of natural regeneration. There is, however, evidence of vole damage to native oak saplings planted in areas cleared of conifers (Truscott et al. 2004). It follows that the common practice of excluding livestock and deer from upland plantations or areas of natural regeneration will result in an increase in vegetation growth (and therefore suitable vole habitat) that in turn will lead to an increase in vole numbers, which may result in damage to newly planted trees.

To date there have been no large-scale experimental manipulations in the uplands that have considered the effects of different grazing regimes on the abundance of field voles. In particular, no study has looked at the intermediate effects of low-intensity grazing on small herbivores in an upland context. Here we present data from such a grazing experiment and demonstrate the immediate effects on field vole populations of modifying grazing regimes. We then discuss the significance of these results for current land-use practices in the uplands where the effects of grazing is an important issue, and consider the development of a range of management options using livestock for maximizing biodiversity in upland habitats.

Methods

study area

The study site was Glen Finglas, Scotland (56°16′N, 4°24′W), a 4085-ha estate grazed by sheep and cattle and typical of many upland areas of Scotland. A replicated, randomized block experiment consisting of six replicates of four treatments was established in 2003, with baseline data collected in 2002. Plots were each approximately 3·3 ha in size.

experimental treatments

During the autumn of 2002, the plots were fenced and the following treatments were randomly allocated to plots within each block: I) conventional stocking rate (3 ewes ha−1); II) one-third conventional stocking rate (sheep); III) one-third conventional stocking rate (sheep and cattle); IV) ungrazed.

During this time one replicate was repositioned in order to reduce variability of vegetation composition and so that all plots occurred on ‘rough grassland’ dominated by Molinia caerulea–Potentilla erecta mire, Juncus effusus/acutiflorus–Galium palustre rush pasture and Festuca oviana–Agrostis capillaris–Galium saxatile grassland communities (Rodwell 1991, 1992).

Within each plot, five points were randomly selected and reused for each recording period. At each point, five quadrats (25 × 25 cm) were searched for the presence and absence of vole signs (Lambin, Petty & MacKinnon 2000). These included runways, fresh clippings and fresh droppings. Surveys were conducted twice a year (April and October) from 2002 to 2004, resulting in 600 quadrats survey−1 (3600 in total). Quadrat locations were randomly reassigned relative to points at each recording period. Vole sign indices have generally been shown to be linearly related to actual vole densities based on snap-trapping methods (Lambin, Petty & MacKinnon 2000) and are widely used to estimate vole abundance.

data analysis

Of the three vole signs recorded per quadrat, we analysed the presence or absence of droppings at the quadrat level as the response variate in the model, as observers were more confident of positively identifying this sign in the field. These data were analysed using a generalized linear mixed model with a binomial error structure and a logit link function using GenStat® 7.2 (VSN International Ltd, Hertfordshire, UK). Random effects were included in the model to reflect the structure of the experimental design (plot, plot by recording period interaction) and data collection (point within plot). Categorical fixed effects in the model reflected the balanced blocking structure (replicate), year, season and interactions between these factors. Also included as fixed effects were categorical variables to account for the grazing regimes allocated to plots for data from 2003 onwards (treatment), the grazing regime that was to be allocated to plots for data from 2002 (pretreatment), and interactions between treatment, year and season. Significance of terms was assessed by referring scaled Wald statistics on adding fixed-effect terms to the model to F-distributions with denominator degrees of freedom determined from the experimental design.

Results

In 2002, the proportion of quadrats that contained vole droppings was similar between plots (Fig. 1) with no significant pretreatment effect (F3,15 = 0·83, P= 0·5; Table 1). However, in 2003 (i.e. commencement of the experiment) the proportion was higher in plots with low stocking densities compared with plots with conventional stocking densities (treatment I) and highest in plots with no livestock grazing (treatment IV). The effect was stronger in 2004 (Fig. 1). The effect of grazing treatment on the proportion of quadrats that contained vole droppings was statistically significant (F3,15 = 11·93, P < 0·001; Table 1). There was a statistically significant effect of year (F2,15 = 17·15, P < 0·001) and season by year (F2,15 = 18·38, P < 0·001; Table 1) on the proportion of quadrats that contained vole droppings, with a year-on-year increase from 2002. We found no other statistically significant effects on the proportion of quadrats that contained vole droppings (Table 1).

Figure 1.

The mean proportion of quadrats (± SE) containing vole droppings at Glen Finglas, Scotland. Treatment I, conventional stocking rate (3 ewes ha−1); treatment II, one-third conventional stocking rate (sheep); treatment III, one-third conventional stocking rate (sheep and cattle); treatment IV, ungrazed. Baseline data for 2002 was collected before the allocation of treatments.

Table 1.  Results of a generalized linear mixed model analysing quadrats containing evidence of fresh vole droppings at Glen Finglas, Scotland. Values in bold are statistically significant
Fixed term*Wd.f.P
  • *

    Fitted sequentially.

  • Scaled Wald statistics.

  • The statistical significance of the scaled Wald statistics was assessed using F-tests with denominator degrees of freedom set to 15 to replicate, treatment and pretreatment and 75 otherwise.

Replicate 1·97 50·14
Season 0·35 10·56
Year17·15 2< 0·001
Season × replicate 1·23 50·30
Year × replicate 1·01100·44
Season × year18·38 2< 0·001
Season × year × replicate 1·67100·10
Treatment11·93 3< 0·001
Pretreatment 0·83 30·50
Year × treatment 0·56 30·64
Season × treatment 0·95 30·42
Season × year × treatment 0·97 30·41

Discussion

We have demonstrated experimentally for the first time that livestock grazing pressure affects the abundance of field voles in upland grassland in Scotland. In the first year of the experiment, immediate treatment effects were detectable with fewer quadrats in the conventionally grazed treatment containing vole droppings compared with those in the ungrazed treatment, but with intermediate vole abundances in the extensively grazed treatments. The significant treatment effects became even more apparent in 2004, with more quadrats containing vole droppings in the extensively grazed mixed treatment than in the extensively grazed treatment that contained only sheep, although the ungrazed treatment still had the highest level of vole activity.

There are two main factors that could explain the relationship between livestock grazing pressure and field vole abundance. First, there is likely to be direct competition for food between small and large herbivores. Where livestock densities are high, the amount of food available to voles would be reduced. Secondly, low livestock densities would result in high, dense vegetation that would have indirect benefits to voles in terms of microclimate (McCafferty, Moncrieff & Taylor 2003) and increased cover from avian predators. According to Smit et al. (2001), small rodents depend on structurally rich vegetation for shelter and large herbivores can have significant effects by reducing the quality of small rodent habitats.

We have demonstrated that field vole abundance is related to grazing pressure. It is, however, the combination of both prey abundance and prey availability that is important for predators such as raptors. Prey availability will be influenced by vegetation structure, thus while vole abundance increases as grazing pressure decreases, the relationship with availability may not be linear but will be highest at intermediate grazing densities.

There was a significant effect of year and year × season on vole abundances. This was expected, as vole population cycles are common in the Palearctic (Gipps & Alibhai 1991; Hanski, Hansson & Henttonen 1991; Lambin, Petty & MacKinnon 2000; Petty et al. 2000; Sherratt et al. 2000). There is considerable interest in understanding the mechanisms that cause the population cycles in field voles (Oli 2003). Despite decades of research and a plethora of hypotheses, population cycles remain one of the most important and unresolved issues in ecology (Krebs 1996; Stenseth 1999; Oli 2003). One of the leading hypotheses of population cycles invokes predation as the underlying mechanism (Hanski, Hansson & Henttonen 1991; Korpimäki & Krebs 1996; Krebs 1996; Hanski et al. 2001). Recent field experiments in northern England have shown that specialist mammalian predators are neither necessary nor sufficient to drive field vole population cycles (Graham & Lambin 2002) and thus the search for the mechanisms underlying vole population cycles continues (Oli 2003). We do not suggest that livestock grazing (or rather the competition between small and large herbivores) could cause cycles, but it could be an important factor influencing vole dynamics and requires further study. In particular, it would be interesting to test whether the complete removal or exclusion of livestock from an area initiates a vole population cycle, and to what extent low grazing pressure might dampen such cycles (see treatment 2; Fig. 1).

In our short-term experiment, the complete removal of livestock led to an immediate and dramatic increase in vole abundance. But the time scale of our study severely limits the predictive power of management recommendations and thus long-term monitoring is needed. Although our experiment was not conducted in a new plantation, it is the exclusion of sheep from land recently enclosed for afforestation that is generally considered to be responsible for the subsequent increase in vole numbers in such areas (Charles 1956). However, long-term studies suggest that after a period of 10–15 years the abundance of field voles in new conifer plantations decreases as the growth of the ground layer vegetation is constrained by canopy shading (Charles 1981). The effects of succession on vole abundance also varies depending on the tree species and growth rate of the trees (Ferns 1979). More research is needed to examine the extent of vole damage to newly planted upland forests and in areas of native woodland regeneration.

The potential differences in priorities between raptor conservationists wishing to increase food abundance for birds of conservation concern and forestry interests (whether commercial or conservation) wishing to protect newly planted trees from vole damage are most likely to occur in newly afforested or areas of natural regeneration where livestock have been excluded. Indeed, it is in such areas where raptor numbers have been shown to increase as they exploit the temporal abundance of small mammals (Lockie 1955; Picozzi & Hewson 1970; Avery & Leslie 1990; Village 1990; Petty 1999).

To limit field vole damage in areas where large herbivores have been excluded, foresters often prescribe the use of split plastic tube guards, which are cheaper and more effective than other methods of protecting broadleaved trees (e.g. hoeing and mowing, chemical repellents). Effective weed control, intended primarily to reduce competition with young trees, also reduces the incidence and severity of vole damage (Davies & Pepper 1993). However, the exclusion of deer and livestock from such areas can dramatically change the structure and composition of the woods (Putman et al. 1989), which may lead to reductions in the diversity and/or abundance of important plant and animal communities (Kirby, Mitchell & Hester 1994).

In order to maximize biodiversity, there have been calls to retain some grazing or browsing in many woodland sites (Kirby, Mitchell & Hester 1994). Although more research is necessary, the regeneration of upland woodlands (such as birch Betula pubescens and Betula pendula) can occur in areas grazed by livestock (Pollock et al. 2005). The challenge is to find innovative and cost-effective ways of incorporating grazing into woodland management on the one hand, whilst limiting the potential amount of tree damage caused by large herbivores on the other. Although specific studies are needed in young plantations, our results nevertheless suggest that low-intensity livestock grazing, used in novel ways, could potentially be a useful management tool to reduce vole abundance (and hence subsequent tree damage) compared with excluding livestock from young plantations completely. Furthermore, vole abundance would still remain higher at low grazing intensity compared with conventional stocking rates, thus still providing more food for raptors. Indeed, low intensity sheep and mixed livestock grazing might improve prey availability as a direct consequence of increased vegetation structure heterogeneity, although more research is necessary to clarify these mechanisms. Where there are advantages in maintaining some grazing to benefit particular groups of species, alternatives to the complete exclusion of large herbivores need to be developed further (Kirby, Mitchell & Hester 1994; Armstrong & Bullock 2003). One alternative is the seasonal use of stock. Depending on the conservation objective, grazing in the winter may benefit early seedling growth of some species by reducing competition from the ground vegetation. However, once the seedlings start to grow above the height of the surrounding vegetation they are vulnerable to winter browsing. A switch to summer or no grazing may then be desirable (Mitchell 1990; Hester, Mitchell & Kirby 1996; Mitchell, Hester & Kirby 1996). Other alternatives might include the use of tethered animals for targeted grazing (i.e. to keep trees out of reach of livestock) or the use of tree guards (which can, however, be costly for large areas) (Hodge & Pepper 1998). Thus there is considerable potential for using large herbivores as a tool in vegetation management in upland woods (reviewed by Hester et al. 2000).

Our results also suggest that it may be possible to maintain the open character of upland habitats and benefit key upland species by reducing sheep grazing pressure and introducing low intensity mixed livestock grazing throughout the uplands. However, although we do not consider game management in this paper, such an approach may not be desirable in areas managed for game birds, especially if it leads to an increase in raptors that might predate on species such as red grouse Lapogus l. scoticus (Thirgood et al. 2000a,b).

Future research should develop innovative ways of bringing low numbers of large herbivores into upland woods whilst minimizing their potential damage to trees. Depending on the biodiversity objective, studies should consider the effects of different grazing management scenarios on tree damage and natural regeneration (reviewed by Kirby, Mitchell & Hester 1994). Furthermore, the extent to which grazing depresses vole population cycles is important, as is understanding the impact of livestock grazing on the availability of voles to raptors and mammalian predators.

Acknowledgements

We thank the Woodland Trust for allowing us to work at Glen Finglas. The research was conducted as part of a collaborative project investigating the effects of grazing management on upland bird populations (GRUB) funded by the Scottish Executive Environment and Rural Affairs Department. We thank the GRUB partners for help and advice. We thank Helen Armstrong, Steve Petty and Mick Marquiss for their help and advice during the preparation of this manuscript.

Ancillary