SEARCH

SEARCH BY CITATION

Keywords:

  • Cervus elaphus;
  • grazing;
  • plant communities;
  • plant diversity;
  • plant productivity

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information
  • 1
    Understanding the effects of red deer Cervus elaphus grazing on a range of plant communities is important for formulating guidelines for the management of forestry plantations, nature reserves and deer hunting enterprises.
  • 2
    The effects of red deer on the species composition of plant communities varying in productivity were examined on the Isle of Rum, Inner Hebrides, north-west Scotland.
  • 3
    The study compared vegetation inside and outside fences erected to exclude deer for more than 20 years, and between areas of the island where different deer culling policies were instigated in 1991.
  • 4
    The effect of the exclusion of deer was directly proportional to vegetation type. Long-term deer exclusion (20–40 years) had negligible effects on the species composition of the most unproductive plant communities, but caused a marked decline in the species richness of productive grassland. The decline was due to the loss of prostrate herb species inside exclosures where Festuca rubra had assumed a high dominance.
  • 5
    Changes in the size and sex ratio of the deer populations were not associated with major changes in plant species richness or relative abundance. The most notable change occurred in productive grassland. Here, Festuca rubra had greater cover and several low-growing herbs such as Polygala serpyllifolia and Thymus polytrichus had lower cover in areas where hind numbers had been reduced than in areas where there was no culling.
  • 6
    High deer densities within an unculled food-limited population were associated with negligible effects on plants in Molinia caerulea flush and Calluna–Molinia wet heath vegetation.
  • 7
    Tree regeneration inside the exclosures or in areas where deer were culled was minor; only two solitary rowan Sorbus aucuparia saplings were found.
  • 8
    Flowerhead densities of herbaceous plant species and grasses showed trends that could be associated with cattle grazing. It is suggested that this may lead to changed recruitment, although seed sowing experiments are required to test this.
  • 9
    The results suggest that red deer grazing sustains the plant species diversity of productive grasslands, with reduced deer grazing leading to the loss of plant species in these communities. In contrast, the effects of reduced deer densities and altered culling policies on unproductive vegetation types are negligible.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

Determining the impact of red deer Cervus elaphus L. grazing on upland plant communities is important for the development of guidelines for the management of forestry plantations, nature reserves and deer-stalking enterprises (Clutton-Brock & Albon 1989), each of which may have different objectives. In forestry plantations or nature reserves where regeneration of trees is required, deer grazing is generally regarded as having a negative effect (Scott et al. 2000), and deer-proof fences or culling to achieve low deer densities are recommended to encourage tree recruitment (Booth 1984; Gong, Swaine & Miller 1991; Scottish Natural Heritage 1994). In contrast, in communities where there is no forestry or woodland to protect, some grazing may be beneficial in maintaining plant species richness. Grazing can help maintain a wide range of species, check the spread of dominant plant species, such as Molinia caerulea, and stimulate grass growth (Ball 1974; Crawley 1997). However, high deer densities may also be detrimental for nature conservation. Grant, Hamilton & Souter (1981) reported that hind densities of 110–250 km−2 reduced Calluna vulgaris cover. In the case of deer-stalking enterprises, the objective of the manager is to maintain large numbers of trophy animals (usually stags), and in many areas this is associated with culling policies that lead to high deer densities (Buckland et al. 1996). Recent modelling studies of Buckland et al. (1996), however, suggest that a sustainable stag harvest may be achievable with much lower overall deer densities (achieved through increased hind culls). Thus, conflicts in management objectives between deer-stalking enterprises and other land uses (e.g. tree regeneration) may be reconciled. However, to date there is little information on how changes in deer densities or sex ratios (through culling) affect the botanical composition of a range of vegetation types.

A number of factors are likely to affect the response of plant communities to deer grazing, including site productivity and the density and sex ratio of deer (Thompson & Miles 1995; Bakker & Berendse 1999; Hester et al. 1999; Ball, Danell & Sunesson 2000). Productivity can determine habitat use by herbivores (Nams, Folkard & Smith 1993; John & Turkington 1997), as well as the capacity of plants to recover from defoliation (Lee et al. 2000). Several studies note that deer prefer productive Agrostis–Festuca grassland over less productive vegetation types (Clutton-Brock & Albon 1989), although stags tend to include a higher proportion of lower quality food in their diet than hinds (Staines, Crisp & Parish 1982).

The objective of this study was to examine how differences in deer culling influenced the botanical composition and vegetation structure of a range of vegetation types on the Isle of Rum, Inner Hebrides, north-west Scotland. The Isle of Rum provides a unique opportunity to investigate how deer management influences the vegetation as areas of the island are assigned to different deer management policies; these include complete grazer exclusion, no culling (i.e. an unregulated food-limited population), equal culling of hinds and stags, and differential culling of stags and hinds to alter sex ratios. The study compared a range of productive (e.g. Agrostis–Festuca grassland) and unproductive (e.g. Calluna heath, Molinia caerulea or Schoenus nigricans fen) vegetation types between areas of the island that had different deer management. Our aim was to test whether changes in deer density and sex ratio brought about by differences in culling policy have measurable short- to medium-term impacts on the species composition, structure and flowering of plant communities.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

DESCRIPTION OF THE ISLAND

The Isle of Rum is a 10 684-ha nature reserve situated in the Inner Hebrides off the north-west coast of Scotland (57°0′N, 6°20′W). Rum has an oceanic climate with mild, wet and windy weather for much of the year. It is a mountainous island, with the highest peak, Askival, rising over 800 m. The productive grasslands characterize valleys near sea shores, while heaths, bogs and fen vegetation prevail on elevated terrain (Ball 1987). The distribution of main vegetation types has been mapped by Ferreira (1970). Clutton-Brock, Guinness & Albon (1982) and Clutton-Brock & Ball (1987) give a description of the flora and fauna on Rum, while Wood (2000) describes the management history of the island. Rum was designated as a National Nature Reserve in 1957.

GRAZERS AND THEIR HISTORY ON RUM

Red deer

Red deer were reintroduced to Rum in 1845 when the bulk of the island's human population left. During the first half of the 20th century the island was developed as a sporting estate for deer hunting, and there was a population of between 1200 and 1700 deer. This was in addition to a flock of about 2000 sheep. Approximately 40 deer of each sex were culled annually during this period. Following the purchase of Rum in 1957 by the Nature Conservancy, the sheep flock was removed, and the deer cull was increased to around one-sixth of the adult population counted in spring, distributed equally between the sexes. In 1972, the island was divided into five deer management blocks (Fig. 1) and culling ceased in block 4 (also known as North Block), but the traditional cull was continued in the other four blocks. These culling treatments continued until 1991, when new treatments were instigated (Clutton-Brock & Thomson 1998; Clutton-Brock et al. 2002): block 1, stag numbers reduced by 50% (annual average cull of 68% of the spring population of males), with hind culls left at their existing level (i.e. one-sixth culled each year); block 2, maintained as a control block with around one-sixth of both sexes culled each year; block 3, hind numbers reduced by 50% (annual cull about 60% in 1991 and 1992, after 1992 14% annual cull), with stag culls left at their existing level (i.e. one-sixth culled each year); block 4, continuation of a no-cull regime; and block 5, mainly high-altitude areas with low densities of deer and annual cull about 10% in 1991–97.

image

Figure 1. Map of the Isle of Rum showing the sampling areas. The boundaries between deer blocks 1–5 are marked with a solid line. The letters A, B, D and E refer to grazing contrasts between deer blocks, with the letter placed between the two blocks being compared. Contrast C was based on grazing exclosures located in block 2 and block 4 (not shown in the map). Contrast F was based on foredune vegetation at Kilmory and Samhnan Insir, which were both located within block 4.

Download figure to PowerPoint

A summary of deer densities in each block is shown in Table 1. These density values show that the culling policy has affected deer densities, even though the reductions in deer numbers have not been directly proportional to the culling treatment. For instance, the relatively small reduction in stag numbers in block 1 has been caused by increased immigration from blocks 2 and 5 (Clutton-Brock et al. 2002).

Table 1.  Areas and red deer densities of the five deer management blocks before and after instigation of the sex ratio cull in 1991. Deer densities (mean values for the given years) and areas according to Clutton-Brock & Thomson (1998) and unpublished data of the Deer Commission for Scotland (1997 spring data). Data for 1996 were not available. Productive grasslands comprise mainly Agrostis–Festuca grassland
BlockArea (km2) YearsDeer density km−2
TotalProductive grasslandsStags per block areaHinds per block areaStags per productive grassland areaHinds per productive grassland area
1Stag numbers reduced by half, one-sixth hinds culled each year18·274·671981–19916·4 8·225·1 32·0
1992–19974·810·018·9 39·0
2One sixth of stags and hinds culled each year21·365·291981–19916·8 6·527·5 26·3
1992–19976·9 6·827.9 27·4
3Hind numbers reduced by half, one-sixth stags culled each year25·073·411981–19916·412·346·9 90·3
1992–19977·6 6·356·0 46·6
4Unculled13·770·731981–19913·910·574·5197·5
1992–19973·0 8·556·7160·3
5Low density red deer, very few culled each year11·001·681981–19911·5 1·5 9·9  9·5
1992–19971·7 2·311·1 14·8
Highland cattle

Highland cattle were introduced to Rum in 1970 in an effort to reverse the decline in plant diversity of the grasslands that had resulted from the removal of the sheep population (Ball 1974). The population size in 1997 was c. 30 (26 cows and four calves; R. Virtanen, unpublished data). The herd grazes in the Harris area (block 2) from late autumn (September) to early summer (June). From mid-summer to early autumn the herd moves to fresh grazing in block 3.

Ponies

A small herd of Highland ponies (c. 10) is kept on the island to provide pack animals for carrying deer carcasses during the hunting season. The herd grazes in the Kilmory area (block 4), and also in the Harris area (block 2) for short periods in the summer.

Goats

A herd of approximately 200 feral goats Capra hircus grazes the island but spends most of its time on the western sea cliffs and adjacent heaths and grasslands (Boyd 1981). The goats feed mainly on heaths (Gordon 1989).

GRAZING CONTRASTS AND VEGETATION SAMPLING

The effect of grazing on the abundance of vascular plant species and bryophytes was examined in several vegetation types (Table 2) chosen to represent the dominant vegetation types on the island and to include a range of levels of productivity. As in any study of compartments within a single island, there is a risk of confounding location effects with treatment effects. Repeated measurements taken within a given treatment and location are spatial pseudoreplicates (sensuHurlbert 1984). To minimize this problem, we carried out an extremely conservative analysis, contrasting pairs of regions with different grazing regimes, only using each data set once. Six contrasts (A–F) were carried out (Table 2). The effects of grazing were assessed in the light of the number of contrasts showing significant differences. Each contrast was decided prior to sampling, under the null hypothesis that there was no difference in the response variable (cover, species richness or inflorescence counts, see below) between the two areas compared. This avoided making multiple post hoc comparisons that could lead to an increased probability of rejecting a true null hypothesis.

Table 2.  Grazing contrasts and vegetation types studied
ContrastVegetation types studiedNVC codeType of contrastGrazing conditions of contrast  
AWet Calluna–Molinia heathH21 M25bAcross fence line erected in mid-1960sBlock 4: high density unculled deer populationvs.Tree plantation area with occasional deer grazing at low density
B1. Wet Calluna–Molinia heathM25b M15Across fence line erected in 1974Block 5: low density red deervs.Tree plantation area with occasional deer grazing at low density
2. Calluna–Racomitriumlanuginosum heathH14    
CProductive (Agrostis–Festuca) grassland and associated grass-rich vegetationU4bLong-term exclosures erected in 1957Grazing exclosures in block 2 and block 4vs.Areas near to enclosures grazed by deer, goats and Highland cattle and ponies
D1. Productive (Agrostis–Festuca) grasslandU4bTwo locations comparedBlock 2, Harris: one-sixth of stags and hinds culled each year, goats, Highland cattle in autumn, winter, springvs.Block 1, Papadil: stag numbers reduced by half, one-sixth hinds culled, goats
2. Herb-rich Calluna heath    
3. Schoenus nigricans fen    
E1. Productive (Agrostis–Festuca) grasslandU4bTwo locations comparedBlock 4, Kilmory: high density unculled deer population, goats, poniesvs.Block 3, Guirdil: hind numbers reduced by half, one-sixth stags culled, Highland cattle in late summer
2. Calluna heathH21    
3. Molinia flushM25b    
F1. Ammophila arenaria foreduneSD6a, SD7dTwo locations comparedBlock 4, Kilmory: high density unculled deer population, poniesvs.Block 4, Samhnan Insir: high density unculled deer population

The vegetation sampling in each contrast was carried out in May 1997, and the flowering surveys in late July 1997. The percentage cover of each species of bryophyte and vascular plant species was recorded in 2 × 2-m quadrats by visual estimates of the projection area of leaves (both dead and green). Cover values greater than 15% were estimated to the nearest 5%. Cover values of less common species (< 10% cover) were assessed in one of eight categories as follows: 0·125%, 0·25%, 0·5%, 1%, 2%, 3%, 5% and 7%. The height of the dominant species (the maximum height of canopy measured from the centre of the quadrat) and the thickness of the moss carpet were recorded for each vegetation type. Nomenclature follows Blockeel & Long (1998) for bryophytes, Stace (1997) for vascular plants and Purvis et al. (1992) for lichens.

DESCRIPTION OF CONTRASTS

Contrast A, north of Kinloch village

This contrast examined Calluna–Molinia wet heath vegetation on either side of the fence line that ran from the northern coast of the island to Kinloch village (Fig. 1 and Table 2). The fence was erected in the mid-1960s (Ball 1987). The area to the west of the fence is part of block 4. The area to the east of the fence is part of a forestry plantation (enclosure) and was supposed to be free of deer. In fact, deer were occasionally observed inside the plantation, so the contrast was considered to be with very low deer densities rather than exclusion. Birch Betula pendula, alder Alnus glutinosa, Scots pine Pinus sylvestris and rowan Sorbus aucuparia were planted closest to the fence line, but these were typically at least 200 m from the fence. The fence line was c. 2·5 km long. Along the fence at even intervals, a total of 20 sites was selected where topography and type of vegetation on either side of the fence were similar. At 16 of these sites, starting 10 m from the fence to avoid edge effects, a 10-m tape was laid out at right angles to the fence. Cover and height measurements were carried out in one 2 × 2-m quadrat that was located on the tape based on a random distance between 0 and 10 m. The 10-m transect method could not be used in the remaining four quadrats located in the northern part of the contrast because the ground fell away too steeply from the fence line. Here, a 4 × 4 grid containing 16 2 × 2-m quadrats was established 4–5 m from the fence, and one 2 × 2-m quadrat located within the grid was sampled, based on the draw of a random number between 1 and 16.

Contrast B, south of Kinloch village

This contrast examined wet Calluna–Molinia heath and Calluna–Racomitrium lanuginosum heath on either side of a fence line that ran south of Kinloch village to the coast (Fig. 1 and Table 2). The fence was erected in 1974. The area to the west and south of the fence is part of block 5. The eastern side of the fence is part of the same forestry plantation (enclosure) described in contrast A. Along the c. 5-km fence line starting from the north-west end, a total of 22 sites, 12 sites representing Calluna–Molinia wet heath and 10 sites representing Calluna–Racomitrium lanuginosum heath, was selected. Cover and height measurements were carried out at each of the 22 sites in a 2 × 2-m quadrat that was chosen using the 10-m line transect method described in contrast A.

Contrast C, long-term grazer exclosures

This contrast examined vegetation inside and outside exclosures erected in 1958 to exclude all large grazers (Ball 1974). Five of the 13 original exclosures (measuring either 5 × 5 m or 10 × 10 m) were sampled. Two of the exclosures were in block 2 (plots 1 and 3 of Ball 1974) and three were in block 4 (plots 12, 13 and 15 of Ball 1974). Plot 1 was in species-poor Agrostis–Festuca grassland, plots 12 and 15 were in species-rich Agrostis–Festuca grassland, and plots 3 and 13 were in vegetation that was intermediate between herb-rich Calluna heath and Agrostis–Festuca grassland. The Harris area is part of block 2 and was also grazed by goats and Highland cattle. The Kilmory area is part of block 4 and was also grazed by goats and ponies (Table 2). Cover and height measurements were carried out in one 2 × 2-m quadrat located inside the exclosure approximately 0·5 m from the southern fence of the exclosure, and also in one 2 × 2-m quadrat located outside the fence approximately 5 m from the southern edge of the exclosure. We were only able to find these five exclosures that represented a more or less productive grassland vegetation type. Some exclosures had collapsed and were not sampled.

Contrast D, block 1 vs. block 2

This contrast examined vegetation between block 2 and block 1 (Fig. 1 and Table 2). Contrasts were carried out in three vegetation types: (i) Agrostis–Festuca grassland (green lawn-like vegetation); (ii) herb-rich Calluna heath adjacent to productive grassland; and (iii) Schoenus nigricans fen. These vegetation types occurred on the slopes of Ruinsval and Àrd Mheann (Ferreira 1970). The Harris area is part of block 2 and Papadil part of block 1. Goats grazed both sites, with Highland cattle grazing in block 2 but not block 1. To select the quadrats for sampling, the vegetation in each area was first surveyed and compared with the vegetation map of Ferreira (1970) to locate the relevant vegetation types. Cover and height measurements were carried out in 10–12 quadrats, each measuring 2 × 2 m, that were randomly selected from each vegetation type in block 2 and block 1. In addition to the vegetation survey carried out in May, flowerheads were counted in late July–early August 1997 in the Agrostis–Festuca grassland vegetation. Flowerheads were counted in 0·4 × 0·4-m quadrats that were selected by throwing quadrats at random as the recorder walked a transect across the vegetation. Flowerheads (single flower or inflorescence) of all vascular plants were counted. In total, 300 quadrats were sampled from block 1 and 400 from block 2.

Contrast E, block 3 vs. block 4

This contrast examined vegetation between block 3 and block 4 (Fig. 1 and Table 2). Contrasts were carried out in three vegetation types: (i) Agrostis–Festuca grassland; (ii) Calluna heath; and (iii) Molinia flush vegetation. The Glen Guirdil-Shellesder area is part of block 3 and had a lower deer density than block 4. Highland cattle grazed in block 3 during late summer (July–August) but not block 4. Goats grazed both areas, although numbers were lower in block 4 than block 3 (Boyd 1981). Ponies also grazed in block 4 (Gordon et al. 1987). Cover and height measurements were carried out in 10–12 quadrats, each measuring 2 × 2 m, in both block 3 and block 4. Inflorescences were counted in late July in 400 quadrats in both block 3 and block 4. Quadrats were selected using the methods described in contrast D.

Contrast F, Kilmory vs. Samhnan Insir

This contrast examined foredune vegetation that was dominated by Ammophila arenaria between Kilmory and Samhnan Insir. Both areas are part of block 4, with the foredunes at Kilmory but not Samhnan Insir accessible to ponies. Cover and height measurements were carried out in 11 quadrats, each measuring 2 × 2 m, in both Kilmory and Samhnan Insir. Quadrats were selected using the methods described in contrast D.

STATISTICAL ANALYSIS

The significance of differences in arcsine-transformed percentage cover for each of the common species, total plant cover and the height of the dominant species was assessed by t-tests for each of the single degree of freedom contrasts. In cases where the plant species were scarce or where their distribution was unbalanced (i.e. species were sporadic or absent in either of the sites), the significance of differences were determined by χ2-tests based on the presence/absence of a species. Species richness (the number of species per 2 × 2-m quadrat) and the number of flowerheads for each of the most common species between block 3 and block 4 (contrast E) and between block 2 and block 1 (contrast D) were analysed using log-linear models with Poisson errors corrected for over-dispersion where appropriate (Crawley 1993).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

CONTRAST A

The effects of 35 years of reduced deer grazing on wet Calluna–Molinia heath were negligible. Total cover, species richness and vegetation height were not significantly different between block 4 and the forestry enclosure (Table 3). The only species that showed a significant difference in cover was Eriophorum vaginatum, which had greater cover in the forestry plantation than block 4.

Table 3.  The effect of reduced deer densities on vegetation height, species richness (species per 4 m2) and percentage cover. Contrast A, forestry enclosure vs. block 4; contrast B, forestry enclosure vs. block 5; contrast C, grazing exclosure vs. nearby grazed areas. All values are means unless specified. Significant difference between either side of contrast: * P < 0·05, ** P < 0·01, *** P < 0·005
 Contrast A wet Calluna heathContrast B Calluna–Molinia heathContrast B Calluna–Racomitrium heathContrast C productive grasslands
EnclosureBlock 4EnclosureBlock 5EnclosureBlock 5ExclosureGrazed
  • χ2-test based on presence/absence data; all other tests of significance were t-tests.

Height of Calluna vulgaris (cm) 33·3 34·0 41·4 28·7 14·312·2***  
Height of grass (cm)       35 2·6***
Height of Molinia caerulea (cm) 14·9 14·5 28·6 25·4 11·6 8·7  
Height of dead Molinia caerulea (cm)   23·2 16·0*  8·6 6·3*  
Thickness of moss carpet (cm)  8·1  6·8 10·7  8·8  9·1 7·0 12·1 0·8*
Median number of species per quadrat 14 14 12 14 1514  833**
Total number of species 35 48 29 36 2931 1951
Total cover1091021141111049910193
Eriophorum vaginatum  2·5  0·2**  0·6  0·3
Festuca rubra  0·01  0·6  0·5 59·018·0*
Hypochaeris radicata 0·7*†
Plantago lanceolata 5·0*†
Plantago maritima 4·6*†
Thymus polytrichus 9·2*†

CONTRAST B

The effects of 33 years of reduced deer grazing on Calluna–Molinia heath and Calluna–Racomitrium lanuginosum heath were negligible. In both vegetation types, the total cover, the cover of individual species and species richness were not significantly different between block 5 and the forestry enclosure (Table 3). In the Calluna–Racomitrium lanuginosum heath, Calluna vulgaris was significantly taller in the forestry enclosure than in block 5. In both the Calluna–Molinia and Calluna–Racomitrium lanuginosum heath, the mulch formed by Molinia caerulea was thicker in the forestry enclosure than in block 5, suggesting greater Molinia offtake from the unfenced ground.

CONTRAST C

The effect of the exclusion of deer and other grazers from productive grasslands was marked. The height of Calluna vulgaris and the thickness of the moss carpet were greater inside the fenced enclosures than in the surrounding grazed areas (Table 3). The cover of Festuca rubra in fenced areas was four times greater than in grazed areas. The frequency of occurrence of Plantago lanceolata, Plantago maritima and Thymus polytrichus was lower in fenced than in grazed areas. Species richness was dramatically reduced in fenced areas compared with grazed areas, and several low-growing species such as Carex flacca, Hypochaeris radicata, Pilosella officinarum, Plantago lanceolata, and Plantago maritima were absent from the fenced areas.

CONTRAST D

In herb-rich Calluna heath, species richness was greater in block 1, where stag numbers had been reduced by half, than in block 2, where one-sixth of stags and hinds were culled each year and where Highland cattle also grazed (Table 4). Total cover and vegetation height were not significantly different between block 1 and block 2. Holcus lanatus, Frullania sp. and Hypnum lacunosum were more abundant in block 1 than in block 2, whereas Calluna vulgaris, Carex pulicaris, Potentilla erecta, Rhinanthus minor and Succisa pratensis were more abundant in block 2 than in block 1.

Table 4.  Height of vegetation layers, species richness per 4 m2, percentage cover and flowerhead densities per 0·16 m2 in contrast D (block 1 vs. block 2). All values are means unless means unless specified. Significant differences indicated as follows: * P < 0·05, ** P < 0·01, *** P < 0·005
 Herb-rich Calluna heathProductive Agrostis–Festuca grasslandSchoenus nigricans fen
CoverCoverFlowerheadsCover
Block 1Block 2Block 1Block 2Block 1Block 2Block 1Block 2
  • χ2-test based on presence/absence data; all other tests of significance were t-tests.

Height of Calluna vulgaris (cm) 2·5 2·6      
Height of grasses (cm) 3·4 3·5 2·4 2·4    
Height of Schoenus nigricans (cm)      48·4 35·1*
Thickness of moss carpet (cm) 1·0 0·7 1·0 0·6   2·3  2·9
Median number of species per quadrat3225*2825  15 14
Number of species per quadrat     5·5 6·0  
Total number of flowerheads per quadrat    20·322·3  
Total number of species62705265  42 41
Total cover93989995  99101
More abundant in block 1
Cynosurus cristatus 2·8 1·7***
Danthonia decumbens 1·4 0·7***
Holcus lanatus, p.p. H. mollis 2·2*† 4·6 0·3* 0·01  0·03
Ranunculus ficaria 0·2 1·4 0·01*
Frullania spp.10·4 2·6*** 0·05
Hypnum lacunosum 3·1 0·3*** 0·9 0·1
Mnium hornum 0·4 0·4*†
More abundant in block 2
Anthoxanthum odoratum 3·8 3·3 4·9 7·7 0·2 0·8*** 4·7  1·8
Bellis perennis 0·6 4·1 6·1 0·2 0·6***
Calluna vulgaris16·732·0** 0·3 1·6 0·2  0·5
Carex flacca, p.p. C. panicea 2·0 4·5 1·6 8·0* 0·3  4·4
Carex pulicaris 3·1**† 2·3*† 0·2 0·2 3·8  0·2
Gentianella campestris 0·03 0·3***
Hypochaeris radicata 0·6 1·9 0·8 1·4 0·1 0·3*** 0·08
Leontodon autumnalis 0·2 0·1 1·5 0·9 0·4 0·2***  0·01
Plantago lanceolata 2·2 2·8 7·1 8·7 4·7 6·3*** 0·4  0·05
Plantago maritima 3·2 4·3 2·6 6·3* 2·5 4·6***
Polygala serpyllifolia 0·1 0·3 0·01 0·1* 0·01
Potentilla erecta 2·3 6·5* 2·9 5·0 0·4 0·5 0·3  0·3
Prunella vulgaris 0·9 0·7 1·3 1·4 0·6 0·8* 0·01  0·01
Rhinanthus minor 0·06*†
Succisa pratensis 0·9 4·6*** 0·5 1·0 0·2  0·3
Trichophorum cespitosum  1·7*†
Trifolium repens 1·1 0·4 5·1 2·6 0·06 0·2***
Viola riviniana 0·7 0·7 0·7 0·1 0·1  0·1
Breutelia chrysocoma 0·2 0·04  0·9*†

In Agrostis–Festuca grassland, species richness, total cover and vegetation height were similar in block 2 and block 1, but some individual species differed significantly in their abundance between sites (Table 4). Carex flacca, Carex pulicaris and Polygala serpyllifolia and were more abundant in block 2 than block 1, whereas Holcus lanatus, Mnium hornum and Ranunculus ficaria were more abundant in block 1 than in block 2. Flowerhead densities of Cynosurus cristatus, Danthonia decumbens and Leontodon autumnalis were significantly higher in block 1 than in block 2, whereas flowerhead densities of Anthoxanthum odoratum, Bellis perennis, Gentianella campestris, Hypochaeris radicata, Plantago lanceolata, Plantago maritima, Prunella vulgaris and Trifolium repens were significantly higher in block 2 than in block 1 (Table 4).

There were few differences in the species composition and structure of Schoenus nigricans fen vegetation between block 2 and block 1 (Table 4). Schoenus nigricans was taller in block 1 than in block 2 but species richness and total cover were the same. Breutelia chrysocoma and Trichophorum cespitosum were more abundant in block 2 than in block 1.

CONTRAST E

There were few differences in the species composition and vegetation structure of low-productivity Calluna heath and Molinia flush vegetation between block 4, where there was an unmanaged deer population and goat and pony grazing, and block 3, where hind numbers were reduced by half and Highland cattle grazed in late summer (Table 5). Total cover, vegetation height and species richness did not differ between block 4 and block 3. The cover of Erica tetralix was significantly greater in Calluna heath in block 4 than block 3. The cover of Hylocomium splendens was significantly greater in Molinia flush vegetation in block 4 than in block 3.

Table 5.  Height of vegetation layers, species richness per 4 m2, percentage cover and flowerhead densities per 0·16 m2 in contrast E (block 3 vs. block 4). All values are means unless specified. Significant difference between block 3 and block 4: * P < 0·05, ** P < 0·01, *** P < 0·005
 Calluna heathMolinia flushProductive Agrostis–Festuca grassland
CoverCoverCoverFlowerheads
Block 3Block 4Block 3Block 4Block 3Block 4Block 3Block 4
  • χ2-test based on presence/absence data; all other tests of significance were t-tests.

Height of Calluna vulgaris (cm) 21·7 20·0      
Height of grass (cm)      4·1  2·6  
Height of Molinia caerulea (cm)  17·7 17·4    
Thickness of moss carpet (cm)  2·8  3·6 4·3  5·8  1·7  1·4  
Median number of species per quadrat 14 1616 15 28 33 4·2 5·0***
Total number of flowerheads per quadrat      13·118·8***
Total number of species 40 4227 29 73 77  
Total cover12412598102108100  
More abundant in block 4
Aira praecox 0 0·6***
Anthoxanthum odoratum  0·2 2·4  1·3  0·9  5·3** 0·4 1·1***
Carex caryophyllea  0·01 0·03  0·3  1·8**
Carex flacca  0·2  0·2 0·06  0·4  0·3  1·9***
Cynosurus cristatus  0·2  0·1 0·6 1·2***
Danthonia decumbens 0·06 1·5***
Erica tetralix  1·0  3·7*** 0·1  0·8
Festuca vivipara 0·7 1·2***
Gentianella campestris 0·08***
Polygala serpyllifolia  0·1  0·04 0·1  0·06  0·02  0·3*
Prunella vulgaris 0·01  0·1  1·1* 0·4 0·2***
Thymus polytrichus  0·2  4·9*† 0·02 0·4***
Dicranum spp.  0·3  2·0 0·06  0·05  0·2  0·8*
Hylocomium splendens  0·7  1·7 0·5  2·7*  7·1  2·4
More abundant in block 3
Bellis perennis  3·9  0·3* 0·4 0·07***
Festuca rubra  0·05 0·4  0·1 26·2 15·8* 0·08 1·2***
Leontodon autumnalis  1·4  0·4 0·4 0·02***
Lolium perenne 0·5 0·06***
Potentilla erecta  0·6  0·3 0·7  0·9  1·7  1·9 0·6 0·4***
Ranunculus acris  0·3  0·4 0·3 0·08***

Differences in Agrostis–Festuca grasslands between block 4 and block 3 were more pronounced (Table 5). Anthoxanthum odoratum, Carex caryophyllea, Polygala serpyllifolia, Prunella vulgaris, Carex flacca, Thymus polytrichus and Hypochaeris radicata were more abundant in block 4 than in block 3, whereas Bellis perennis and Festuca rubra were more abundant in block 3 than in block 4. There were marked differences in flowerhead densities between block 3 and block 4. One grass species, Lolium perenne, and five herb species, Bellis perennis, Leontodon autumnalis, Potentilla erecta, Prunella vulgaris and Ranunculus acris, had higher flowerhead densities in block 3 than in block 4. Six grass species, Aira praecox, Anthoxanthum odoratum, Cynosurus cristatus, Danthonia decumbens, Festuca rubra and Festuca vivipara, one herb species, Gentianella campestris, and a prostrate dwarf shrub, Thymus polytrichus, had higher flowerhead densities in block 4 than in block 3.

CONTRAST F

There were marked differences in foredune vegetation between Samhnan Insir, which is grazed only by deer, and Kilmory, which is grazed by both deer and ponies (Table 6). Total cover, species richness and the thickness of the moss carpet were all greater at Samhnan Insir than at Kilmory. The cover of Ammophila arenaria, Cardamine hirsuta, Hylocomium splendens and Rhytidiadelphus squarrosus and the frequency of occurrence of Sedum acre were greater at Samhnan Insir than at Kilmory. The cover of Hypochaeris radicata and the frequency of occurrence of Plantago lanceolata and Tortula ruraliformis were greater at Kilmory than at Samhnan Insir.

Table 6.  Height of vegetation layers, species richness per 4 m2 and percentage cover of Ammophila arenaria dune vegetation in contrast F (Samhnan Insir vs. Kilmory). All values are means unless specified. Significant difference between Samhnan Insir and Kilmory: * P < 0·05, ** P < 0·01, *** P < 0·005
 Samhnan InsirKilmory
  • χ2-test based on presence/absence data; all other tests of significance were t-tests.

Height of Ammophila arenaria (cm)37·926·7
Thickness of moss carpet (cm) 2·8 1·3*
Median number of species per quadrat2013*
Total number of species4029
Total cover9366*
More abundant in Samhnan Insir
Ammophila arenaria58·530·5***
Cardamine hirsuta 0·4 0·05***
Sedum anglicum 0·2*†
Hylocomium splendens 5·2 0·1*
Rhytidiadelphus squarrosus13·5 3·1**
More abundant in Kilmory
Plantago lanceolata 0·2 1·5*
Tortula ruraliformis 0·810·3***

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

LIMITATIONS OF THE APPROACH

In any observational study such as this, where experimental treatments are not randomly allocated to initially homogeneous material, there is always the risk that differences between contrasts are caused by confounding variables rather than by the hypothesized mechanism (difference in deer grazing pressure in our case). Any one of the contrasts is potentially confounded by countless site factors such as soil and microclimate (although we attempted to control for gross effects like slope, aspect and geology at the planning stage, when the contrasts were selected). Also, there will be varying proportional impacts of different herbivores (deer, cattle, ponies, goats) in different places. It is only through the replication that the study gains any statistical power. The fact that the contrasts tell a consistent story lends weight to the hypothesis that the observed differences can be interpreted as effects of differences in deer grazing. Likewise, where we do not observe consistent differences, the replication encourages us to accept the hypothesis that deer grazing does not have major impacts on these particular plant communities over the time scales measured. As Hairston (1989) stresses, however, the results of observational studies can never be more than hypotheses until such time as they are assessed by controlled experiment.

If the contrasts are wholly independent, then the increase in power is greatest. Lots of contrasts on the same island (as here) could always be subject to the criticism that (at best) our results apply only to the situation on Rum. While these concerns are valid, the alternative is to ignore information from natural experiments altogether, and to insist on manipulative experiments for all ecological questions. Students of evolution would question this, and so would most practising field ecologists. Manipulative experiments are no panacea, of course, because they often set in train profound transient dynamics, but what we are really interested in is the long-term response (and this may be quite different from the short-term transient response).

EFFECTS OF GRAZER EXCLUSION

The key finding of this study is the importance of plant productivity in assessing the impact of deer grazing on plant community structure. In the highest productivity grasslands (mainly Agrostis–Festuca‘greens’), vascular plant species diversity was four times higher in grazed than in ungrazed areas, and deer reduced standing biomass to a fraction of the level that was found within exclosures. In contrast, in the least productive wet heaths, there was very little evidence for any significant impact of deer grazing on plant species diversity or abundance.

In keeping with previous studies (Kydd 1964; Bullock et al. 2001), the species that were lost inside the exclosures situated on productive grassland were the rosette-forming (e.g. Hypochaeris radicata, Pilosella officinarum and Plantago lanceolata) and sprawling, prostrate (e.g. Thymus polytrichus) taxa. The most likely explanation for the loss of these species is increased competition for light and nutrients within exclosures arising from the accumulation of the ungrazed biomass of the perennial grass Festuca rubra.

The small impact of deer grazing on the wet heaths probably reflects low offtake from these communities. It is extremely difficult to obtain accurate measures of offtake in plant communities such as the wet heaths of Rum, where spatial heterogeneity in standing crop biomass is much greater than the annual variation in biomass through production and offtake at a particular point in space (Tuke 2001; M.J. Crawley, unpublished data). However, previous studies show that red deer strongly prefer productive grasslands (Clutton-Brock, Guinness & Albon 1982; Gordon 1989) and use them much more frequently than Calluna moorland or Molinia grasslands (Clutton-Brock & Albon 1989). Thus, it is likely that offtake is not large, and hence the difference in plant community structure between grazed and ungrazed areas is negligible.

There was little tree recruitment within exclosures in any vegetation type, probably due to seed limitation (low seed production and ineffectual dispersal) or low recruitment within dense perennial grass vegetation. Several studies have shown a negative impact of ground cover and vegetation height on tree regeneration (Gong, Swaine & Miller 1991; Ohlson & Zackrisson 1992; Scott et al. 2000). Indeed, some studies demonstrate that grazing and trampling may promote regeneration by creating competition-free microsites for seedling establishment (Miles & Kinnaird 1979; Scott et al. 2000).

EFFECTS OF DEER CULLING

Significant impacts of the increased hind cull were restricted to the productive grasslands. Here, species richness was unaffected but (as for grazing exclosures) several low growing herb species were less abundant and Festuca rubra was more abundant in the culled (block 3) than in the unculled (block 4) population. It is difficult, however, to assign these differences entirely to the difference in deer density brought about by the increased hind cull. The result may be due to reduced deer densities in block 3 relative to block 4 since 1972 (particularly per unit area of productive grassland) rather than to a sex ratio difference in the cull in block 3 since 1991 (Table 1). It is clear, however, that there was little evidence that intense grazing by an unculled (food limited) population had caused botanical impoverishment of these grasslands, i.e. overgrazing (Crawley 1997).

There was also little evidence of significant impacts of the increased stag cull in 1991 on plant species abundance in any of the studied community types (Table 4). It is possible that stag densities were not reduced substantially by the new culling policy, and/or that hind numbers increased when the stag cull was instigated (Clutton-Brock et al. 2002). Indeed, the increased hind density may be an explanation of the lower Calluna vulgaris cover where the stag cull was increased. Past studies note that Calluna vulgaris cover decreases as deer density increases (Grant, Hamilton & Souter 1981). Although there was little impact of stag culling on plant cover in productive grassland, flowerhead densities were affected, with seven herb species having more flowerheads where the stag cull had not been increased. The exact reason for this is unclear, but it may reflect the movement of cattle away from this block in late summer, which allowed flowerheads to escape defoliation, so leading to insignificant differences in cover being expressed as increased flowerhead densities (e.g. Plantago lanceolata, Prunella vulgaris and Trifolium repens). We did not measure seedling recruitment in order to ascertain whether differences in flowering (and so seed production) would affect species diversity. However, we know from previous work that several of the species that showed increased flowerhead densities (e.g. Plantago lanceolata and Trifolium repens; Edwards & Crawley 1999a,b) can be seed limited. This leads to the expectation that changed production of propagules may affect species diversity in the longer term (Turnbull, Crawley & Rees 2000).

PONIES AND FOREDUNE VEGETATION

The single contrast conducted in foredune vegetation provided tentative evidence that the presence of ponies in dunes reduced the abundance of the dominant Ammophila arenaria and decreased species richness. Pony usage of dunes is low (percentage of feeding sightings 3·7%, 9·1%, 0·1% and 21·4% for winter, spring, summer and autumn, respectively; Gordon 1989), which suggests they may have a small effect. However, our personal observations (R. Virtanen, G.R. Edwards & M.J. Crawley) revealed that ponies caused damage to the foredune, with their hooves uprooting plants and breaking up patches of the dominant Ammophila arenaria. Similar factorial experiments to those conducted by Hewett (1985) would further clarify the effects of ponies on dune vegetation on Rum.

MANAGEMENT IMPLICATIONS

One of the concerns associated with the exclusion of herbivores or reduced herbivore densities in upland plant communities is a reduction in plant species richness and increased dominance of a few species (Ball 1974; Wood 2000). Our study indicates that this is a valid concern in productive communities (e.g. Agrostis–Festuca grasslands) but of less importance in unproductive communities (e.g. Calluna–Molinia wet heath), where effects of exclusion are small and/or slow. Efforts to restore unproductive communities following extreme grazing events should take account of the fact that the effects can be small and restoration may take long periods (> 20 years; Lee et al. 2000).

There has also been concern that very high grazing pressure by deer or other domestic livestock can have negative impacts on plant biodiversity. However, in our study, the high deer densities of an unculled deer population in block 4 were not associated with negative impacts on plant biodiversity. Culling policies that alter the density and sex-specificity of the deer populations did not cause major changes in species richness or relative abundance, at least in the short term (7 years). Finally, in our study none of the herbivore exclusion or culling policies had any measurable effect on tree regeneration. To understand the relationship between deer density and tree regeneration will require experimental work in which seed inputs, soil disturbance and plant competition are manipulated in order to tease apart the relative importance of seed limitation, microsite limitation and competitive exclusion. Ideally, such an experimental protocol would be crossed with a factorial combination of deer and other herbivore densities.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

We thank: staff of Scottish Natural Heritage; Martin Curry and Ann Taylor for help in transportation and accommodation; Tim Clutton-Brock for lodging in Kilmory research station; the Deer Commission for Scotland and Derek Thomson for information on red deer populations; Dr E.J.P. Marshall and three anonymous referees for constructive comments. The Finnish Research Council of Natural Resources and Environment (R. Virtanen) and the UK Natural Environment Research Council (G. R. Edwards, M. J. Crawley) supported the study financially.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information
  • Bakker, J.P. & Berendse, F. (1999) Constraints in the restoration of ecological diversity in grassland and heathland communities. Trends in Ecology and Evolution, 14, 6368.
  • Ball, M.E. (1974) Floristic changes on grasslands and heaths on the Isle of Rhum after a reduction of exclusion of grazing. Journal of Environmental Management, 2, 299318.
  • Ball, M.E. (1987) Botany, woodland and forestry. Rhum: The Natural History of an Island (eds T.H.Clutton-Brock & M.E.Ball), pp. 4362. Edinburgh University Press, Edinburgh, UK.
  • Ball, J.P., Danell, K. & Sunesson, P. (2000) Response of a herbivore community to increased food quality and quantity: an experiment with nitrogen fertilizer in the boreal forest. Journal of Applied Ecology, 37, 247255.
  • Blockeel, T.L. & Long, D.G. (1998) A Check-List and Census Catalogue of British and Irish Bryophytes. British Bryological Society, Cardiff, UK.
  • Booth, T.C. (1984) Natural regeneration in the native pinewoods of Scotland: a review of principles and practice. Scottish Forestry, 38, 3342.
  • Boyd, I.L. (1981) Population changes and the distribution of a herd of feral goats (Capra sp.) on Rhum, Inner Hebrides, 1960–1978. Journal of Zoology, London, 193, 287304.
  • Buckland, S.T., Ahmad, S., Staines, B.W., Gordon, I.J. & Youngson, R.W. (1996) Estimating the minimum population size that allows a given annual number of mature red deer stags to be culled sustainably. Journal of Applied Ecology, 33, 118130.
  • Bullock, J.A., Franklin, J., Stevenson, M.J., Silvertown, J., Coulson, S.J., Gregory, S.J. & Tofts, R. (2001) A plant trait analysis of responses to grazing in a long-term experiment. Journal of Applied Ecology, 38, 253267.
  • Clutton-Brock, T.H. & Albon, S.D. (1989) Red Deer in the Highlands. BSP Professional Books, Oxford, UK.
  • Clutton-Brock, T.H. & Ball, M.E. (1987) Rhum: The Natural History of an Island. Edinburgh University Press, Edinburgh, UK.
  • Clutton-Brock, T.H. & Thomson, D. (1998) Manipulation of Red Deer Density on Rum: Synthesis 1992–1995. Scottish Natural Heritage Research, Survey and Monitoring Report No. 97.
  • Clutton-Brock, T.H., Coulson, T.N., Milner-Gulland, E.J., Thomson, D. & Armstrong, H.M. (2002) Sex differences in emigration and mortality affect optimal management of deer populations. Nature, 415, 633637.
  • Clutton-Brock, T.H., Guinness, F.E. & Albon, S.D. (1982) Red Deer: Behaviour and Ecology of Two Sexes. Edinburgh University Press, Edinburgh, UK.
  • Crawley, M.J. (1993) GLIM for Ecologists. Blackwell Science, Oxford, UK.
  • Crawley, M.J. (1997) Plant–herbivore dynamics. Plant Ecology (ed. M. J.Crawley), pp. 401474. Blackwell Science, Oxford, UK.
  • Edwards, G.R. & Crawley, M.J. (1999a) Rodent seed predation and seedling recruitment in mesic grassland. Oecologia, 118, 288296.
  • Edwards, G.R. & Crawley, M.J. (1999b) Effects of disturbance and rabbit grazing on seedling recruitment of six mesic grassland species. Seed Science Research, 9, 145156.
  • Ferreira, R.E.C. (1970) Vegetation Map of the Isle of Rhum. Nature Conservancy, London, UK.
  • Gong, Y.L., Swaine, M.D. & Miller, H.G. (1991) Effects of fencing and ground preparation on natural regeneration of native pinewood over 12 years in Glen Tanar, Aberdeenshire. Forestry, 64, 157168.
  • Gordon, I.J. (1989) Vegetation community selection by ungulates on the Isle of Rhum. II. Vegetation community selection. Journal of Applied Ecology, 26, 5364.
  • Gordon, I.J., Dunbar, R., Buckland, D. & Miller, D. (1987) Ponies, cattle and goats. Rhum: The Natural History of an Island (eds T.H.Clutton-Brock & M.E.Ball), pp. 110124. Edinburgh University Press, Edinburgh, UK.
  • Grant, S.A., Hamilton, W.J. & Souter, C. (1981) The responses of heather-dominated vegetation in north-east Scotland to grazing by red deer. Journal of Ecology, 69, 189204.
  • Hairston, N.G. (1989) Ecological Experiments: Purpose, Design and Execution. Cambridge University Press, Cambridge, UK.
  • Hester, A.J., Gordon, I.J., Baillie, G.J. & Tappin, E. (1999) Foraging behaviour of sheep and red deer within natural heather grass mosaics. Journal of Applied Ecology, 36, 133146.
  • Hewett, D.G. (1985) Grazing and mowing as management tools on dunes. Vegetatio, 62, 441447.
  • Hurlbert, S.H. (1984) Pseudoreplication and the design of ecological field experiments. Ecological Monographs, 54, 187211.
  • John, E. & Turkington, R. (1997) A 5-year study of the effects of nutrient availability and herbivory on two boreal forest herbs. Journal of Ecology, 85, 419430.
  • Kydd, D.D. (1964) The effect of different systems of cattle grazing on the botanical composition of permanent downland pasture. Journal of Ecology, 52, 139149.
  • Lee, W.G., Fenner, M., Loughnan, A. & Lloyd, K.M. (2000) Long-term effects of defoliation: incomplete recovery of a New Zealand alpine tussock grass, Chionochloa pallens, after 20 years. Journal of Applied Ecology, 37, 348355.
  • Miles, J. & Kinnaird, J.W. (1979) Grazing: with particular reference to birch, juniper and Scots pine in the Scottish Highlands. Scottish Forestry, 33, 280289.
  • Nams, V.O., Folkard, N.F.G. & Smith, J.M.N. (1993) Effects of nitrogen fertilization on several woody and non-woody boreal species. Canadian Journal of Botany, 71, 9397.
  • Ohlson, M. & Zackrisson, O. (1992) Tree establishment and microhabitat relationships in north Swedish peatlands. Canadian Journal of Forest Research, 22, 18691877.
  • Purvis, O.W., Coppins, B.J., Hawksworth, D.L., James, P.W. & Moore, D.M. (1992) The Lichen Flora of Great Britain and Ireland. Natural History Museum Publications, London, UK.
  • Rodwell, J. (1991) British Plant Communities, Vol. 2. Mires and Heaths. Cambridge University Press, Cambridge, UK.
  • Rodwell, J. (1992) British Plant Communities, Vol. 3. Grasslands and Montane Communities. Cambridge University Press, Cambridge, UK.
  • Scott, D., Welch, D., Thurlow, M. & Elston, D.A. (2000) Regeneration of Pinus sylvestris in a natural pinewood in NE Scotland following reduction in grazing by Cervus elaphus. Forest Ecology and Management, 130, 199211.
  • Scottish Natural Heritage (1994) Red Deer and the Natural Heritage. SNH Policy Paper. SNH Publications and Graphics, Edinburgh, UK.
  • Stace, C. (1997) New Flora of the British Isles, 2nd edn. Cambridge University Press, Cambridge, UK.
  • Staines, B.W., Crisp, J.M. & Parish, T. (1982) Differences in the quality of food eaten by red deer Cervus elaphus stags and hinds in winter. Journal of Applied Ecology, 19, 6577.
  • Thompson, D.B.A. & Miles, J. (1995) Heaths and moorlands: some conclusions and questions about environmental change. Heaths and Moorland: Cultural Landscape (eds D.B.A.Thompson, A.J.Hester & M.B.Usher), pp. 362385. HMSO, Edinburgh, UK.
  • Tuke, A. (2001) Grass production and sheep grazing on the island of Hirta, Scotland. PhD Thesis. University of London, London, UK.
  • Turnbull, L., Crawley, M.J. & Rees, M. (2000) Are plant populations seed-limited? A review of seed sowing experiments. Oikos, 88, 225238.
  • Wood, B. (2000) Room for nature? Conservation management of the Isle of Rum, UK, and prospects for large protected areas in Europe. Biological Conservation, 94, 93105.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

Table S3. The effect of reduced deer densities on vegetation height, species richness (species per 4 m2) and percentage cover. Table S4. Height of vegetation layers, species richness per 4 m2, percentage cover and flowerhead densities per 0?16 m2 in contrast D (block 1 vs. block 2). Table S5. Height of vegetation layers, species richness per 4 m2, percentage cover and flowerhead densities per 0?16 m2 in contrast E (block 3 vs. block 4). Table S6. Height of vegetation layers, species richness per 4 m2 and percentage cover of Ammophila arenaria dune vegetation in contrast F (Samhnan Insir vs. Kilmory).

FilenameFormatSizeDescription
JPE_734_sm_tables.doc317KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.