SEARCH

SEARCH BY CITATION

Keywords:

  • Calluna vulgaris;
  • grazing experiments;
  • heather utilization;
  • residual maximum likelihood

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Many habitats of high conservation value are managed by grazing but could be damaged by poor grazing management. Approaches and methods to set appropriate grazing regimes must be developed that can be applied with confidence under different situations to ensure that deleterious habitat changes are unlikely to occur.
  • 2
    Heather moorland is an important habitat for conservation, but is a cultural plagio-climax with low productivity and is under threat from high grazing pressures. Too high a pressure converts the dwarf-shrub dominance to grass dominance. Thus, it is imperative to be able to assess a sustainable grazing level for moorland: the no-effect level.
  • 3
    Data from ten grazing experiments on heather moorland, each carried out and monitored in a similar manner, were analysed together to estimate the impact of heather utilization on the balance between heather and grasses, sedges and rushes.
  • 4
    The analysis indicated a no-effect level of 31·6% utilization of current year's growth to maintain the balance between heather and monocotyledonous plants. However, the 95% confidence intervals for the fitted line crossed zero change at 22·5% and 41·4% utilization; a considerable degree of uncertainty that indicates a precautionary approach would be appropriate.
  • 5
    Synthesis and applications. The current assumed utilization level for sustainable grazing of heather (40% of current year's growth) appears too high. A conservative utilization figure to reduce the risk of heather loss should be set nearer 20%. Developing models based on utilization is more efficient than basing models on stocking rate information due to the wide range of factorial combinations of stocking rate and vegetation composition that would need investigating. This approach could be extended to other vegetation types where monitoring of key indicator species could be more efficient than developing experimentally based grazing prescriptions.

Introduction

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

Grazing management is used to derive economic returns from land not suitable for more intensive agriculture and to manage semi-natural habitats for biodiversity benefits. However, poor management, whether it be overgrazing or undergrazing, can result in damage that can alter the long-term economic return or impact on the biodiversity present in the system (Hanley et al. 2008). To prevent this, approaches and methods need to be developed that ensure that deleterious habitat changes do not occur. To be efficient, these approaches need to be robust across different situations, and hence should not need to be developed de novo for each different area or vegetation type.

Heather moorland, also known as upland heathland, has been recognized as a habitat of high conservation value by both UK and European designations [e.g. Annex 1 habitat in the EC Habitats Directive (92/43/EEC) and UK BAP Priority Habitat Action Plans (UKBAP 2007)]. This value is mainly associated with the dwarf shrub and bryophyte flora, the rich invertebrate assemblage, and the breeding and feeding bird assemblages that are associated with high, but not complete, covers of dwarf shrubs (Thompson et al. 1995). Large areas of heathland habitat were lost over the course of the 20th century across many countries of Western Europe. In the lowlands, much of this loss was to agricultural improvement, although some was also lost to succession (Farrell 1989; Mackey, Shewry & Tudor 1998). However, within the uplands, especially in the UK, these losses have been a result of conversion to forestry or a result of high grazing levels that have shifted the competitive balance in favour of grasses and away from heather Calluna vulgaris (L.) Hull and other ericoid shrubs (Thompson et al. 1995; Nolan et al. 2002). This ‘overgrazing’ equates to the ‘conservation’ and ‘wildlife’ overgrazing categories of Mysterud (2006); where in the former the manager is concerned with multiple species and in the latter with maintaining high populations of game species, in particular red grouse Lagopus lagopus scoticus Latham. From a ‘range management’ perspective, high levels of grazing shift the system into a more productive state due to the dominance of grasses, but there is an alteration in the seasonality of food availability, especially where evergreen heather is replaced by deciduous Molinia caerulea (L.) Moench. Nitrogen deposition has increased the rate of this conversion in many areas (Bobbink, Hornung & Roelofs 1998). High grazing levels have also resulted in much of the remaining upland heather being in poor condition (Bardgett, Marsden & Howard 1995; JNCC 2007). Consequently, to conserve and enhance the remaining heather moorland resource, a number of different agri-environment and conservation schemes have been implemented within the UK.

The prescriptions set out within these schemes to limit grazing intensity have largely focussed on setting maximum stocking levels for livestock [e.g. England's Environmental Stewardship (Defra 2005), Northern Ireland's Environmentally Sensitive Areas Scheme and Countryside Management Scheme (DARD 2007) and Wales’ Tir Gofal (Welsh Assembly Government 2007)], or are based on adjusting stocking levels such as Scotland's Rural Stewardship Scheme (The Scottish Government 2007). This method of setting ‘sustainable’ stocking rates has the benefit of being easily understandable and easily audited. However, it ignores the fact that the impact of grazing is dependent on the proportion of the current year's growth that is removed (utilization) from the heather. This utilization is a function of the stocking level, the proportion of heather relative to other vegetation (and especially the more preferred grasses), the available biomass for grazing and the respective preferences of the grazer for the different species present (Grant et al. 1982; Armstrong et al. 1997a). For a set stocking rate, if grass is at a higher proportion in the vegetation, then the utilization of heather is lower, and if below a certain threshold of utilization, the heather can increase at the expense of grass (Hulme et al. 2002; Pakeman et al. 2003). If there is a high cover of heather, then the same stocking density could produce a high degree of heather utilization and, hence, a decrease in cover to the benefit of grass. Consequently, attempts to set levels of stocking for heathland/moorland management are overly simplistic; they may still be set too high and allow damage under certain circumstances or they may prevent appropriate economic utilization of these habitats when there is no risk of damage. Monitoring the proportion of heather shoots grazed (Welch 1984; Welch & Scott 1995) has been suggested and used as an alternative means of setting or monitoring sustainable grazing levels (ADAS 1998; Kirkham et al. 2005). This ‘Grazing Index’ has been shown to be well correlated to measured utilization at low utilization rates (< 40%) where sheep grazing only was occurring (Armstrong & Macdonald 1992). However, use of this index could underestimate damage at high utilization rates, as there may be repeated grazing of the same shoot (Armstrong & Macdonald 1992) and, as different grazers have different bite depths, then it is unlikely to be informative where different herbivores are grazing.

An alternative means of setting ‘sustainable’ grazing levels is through quantifying a level of utilization – measured as the percentage of current-year growth removed – that can be tolerated by the heather. This approach allows comparisons between different herbivore species and across different ecosystems (Bilyeu, Cooper & Hobbs 2007). Attempts to identify the utilization level that conserves or enhances the cover of heather have been based on physiological (Read, Birch & Milne 2002) or mechanistic models (Armstrong et al. 1997a,b), single experiments (e.g. Grant et al. 1982; Hulme et al. 2002; Pakeman et al. 2003), amalgamations of different data types (Palmer et al. 2004) or expert judgement from survey and experimental evidence (MacDonald et al. 1998). However, they are all ultimately based on restricted data sets and, hence, might have limited application.

Setting sustainable grazing levels as a function of utilization may represent a suitable alternative approach to setting management goals, but further information is needed on their operation. The objectives of this study were, firstly, to analyse data from ten controlled, grazing experiments on heather moorland to identify a utilization level below which heather should increase in cover at the expense of monocotyledonous plants (grasses, sedges and rushes), and to calculate the levels of uncertainty relating to this relationship. Secondarily, our aim was to compare these findings to those generated from analysing the data based on stocking rates. The development of methods to set sustainable grazing levels on the basis of impacts to grazing-sensitive species as opposed to setting stocking levels should allow a better fit of management to different circumstances – preventing damage on sensitive sites and allowing appropriate economic utilization of sites where the sensitive species are not heavily utilized.

Material and methods

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

available experiments

Data from ten experiments spanning three regions of the UK (north-east England, north-east Scotland and western Scotland) carried out by the Hill Farming Research Organisation and the Macaulay Institute were available (Table 1; Supporting Information, Appendix S1). These experiments shared common features in that heather utilization had been recorded directly using the method of Grant, Hamilton & Souter (1981), which estimates the removal of shoot material as a percentage of the current year's total growth, rather than as the proportion of shoots grazed or as a proxy such as dung counts (e.g. Welch & Scott 1995). Grant et al. (1981) used the following method to assess the proportion of current year's growth removed by grazing. Each selected shoot is classified into four categories by comparison with ungrazed shoots on the same plant: 0, no grazing; 1, removal of < 50%; 2, removal of 50–100%; 3, removal of 100% of the current year's growth and part of the previous year's growth. Category 3 represents grazing into the previous year's wood. The proportion of current year's growth removed is then calculated as utilization (%) = [(0·3 × no. shoots in category 1) + (0·8 × no. shoots in cat. 2) + (1·2 × no. shoots in cat 3)]/(100 × no. of shoots measured). Typically, ungrazed, new shoots are 2–4 cm in length but extension varies depending on climate and soil fertility. Although growth form can be severely affected by continued heavy grazing, heather plants still produce new growth in the same way and hence it is possible to estimate utilization at all stages and on all growth forms of heather (MacDonald et al. 1998).

Table 1.  Summary of the experiments used in the analysis
ExperimentNational grid referenceSheep density (range, no. ha−1 year−1)Heather phase, and initial heather cover (proportion ± 1 SE)Source
Claonaig matureNR 8705870–1·5Mature, 0·62 ± 0·039Unpublished
Claonaig regenerationNR 8585560–1·2Pioneer, 0·47 ± 0·031Unpublished
DufftownNJ 3633880–1·9Mature, 0·45 ± 0·029Pakeman et al. (2003)
DundonnellNH 0729140–1·2Pioneer, 0·31 ± 0·015Unpublished
GlensaughNO 665 7950–6·0Mature, 0·94 ± 0·005Grant et al. (1982)
OtterburnNT 915019Not measuredMature, 0·48 ± 0·087Unpublished
Redesdale – Burnhead matureNY 8229450–1·8Mature, 0·39 ± 0·015Unpublished
Redesdale – Burnhead regeneration 1NY 8229450–1·2Pioneer, 0·12 ± 0·012Unpublished
Redesdale – Burnhead regeneration 2NY 8229450–1·8Pioneer, 0·09 ± 0·016Unpublished
Redesdale – Road cutNY 8299180–2·1Mature, 0·50 ± 0·016Hulme et al. (2002)

Utilization estimates were made in late winter/early spring to ensure the impact of winter grazing was covered by the measurements, and plot estimates were calculated from at least 100 randomly selected shoots. Sheep densities had been managed and recorded (except for the experiment at Otterburn), the vegetation had been monitored throughout the course of the experiment and each experiment lasted 5 or 6 years. They also covered a wide range of situations in terms of original heather cover, plant community composition and heather growth phase, and each had multiple treatments spanning a range of sheep densities and seasons of grazing (Table 1; Supporting Information, Appendix S1).

Using the rate of change of the proportion of heather in the vegetation could be seen as an oversimplification of the data. It does not, for instance, take into account the overall productivity of an area of moorland, nor the difference in preference between heather and the different moorland grasses. However, it should be a robust metric as it can account for the behaviour of moorland under the following possibilities: (i) reduced grazing leading to an increase in heather at the expense of, or faster than, an increase in grasses and other monocotyledons will result in positive values for this metric; and (ii) increased grazing leading to reduced heather as grasses predominate will result in negative values. Other possible scenarios are also covered, such as: (iii) reduced grazing leading to an increase in grasses faster than an increase in heather would have a negative value, which would be appropriate, given the long-term impacts of grass growth on heather regeneration and the potential for tall grasses such as Molinia caerulea to shade out heather (Pakeman & Marshall 1997; Hulme et al. 2002); and (iv) increased grazing with the proportion of grass decreasing is unlikely as heather is much less tolerant of repeated defoliation and trampling than grasses (Bayfield 1979; Hester & Baillie 1998).

statistical analysis

To address the aim ‘identification of the utilization level below which heather should increase at the expense of monocotyledonous plants’ point quadrat-derived data (total hits) from each plot (74 in total) were first converted to a proportion of heather in the vegetation. Linear regression was then used to produce a rate of change of the proportion of heather in the vegetation against time. This rate of change was then fitted against mean heather utilization or sheep stocking density using a linear mixed-effect model with residual maximum likelihood as the fit criterion (REML, GenStat, Lawes Agricultural Trust 2007). Mean heather utilization/stocking density was included as a fixed effect, and block nested within site as the random effect. Additional analyses which tested the effects of utilization2 or stocking rate2, region, winter/summer grazing, dry/wet heath and heather growth phase as fixed effects were also carried out to assess if they were also significant. Variables were added in a stepwise manner with the condition of inclusion being a significant reduction in deviance. The percentage of variance explained within each stratum of the analysis and the 95% confidence intervals were calculated.

Results

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

There was a clear relationship between the rate of change in the proportion of heather and its utilization (Wald/d.f. = 102·5, P < 0·001, Fig. 1a) Parameter estimates are shown in Supporting Information, Appendix S2. This explained 70·6% of the variance at the site level and 8·1% at the within-site level. Additional terms tried in the model – utilization2, summer/winter, region and phase – were not significant at P < 0·05 and did not reduce the deviance significantly. The derived relationship predicted that there would be no change in heather cover at 31·6% utilization of current season's shoots. However, the 95% confidence intervals for the fitted line cross zero change at 22·5% and 41·4%, indicating that there was considerable spread within the data.

image

Figure 1. Rate of change of heather proportion as a function of (a) utilization (%) or (b) stocking rate (sheep ha−1 year−1), and (c) relationship between stocking rate and utilization. — represents the fitted line from Residual Maximum Likelihood and -- the 95% confidence intervals. Site symbols are: inline image, Claonaig mature; inline image, Claonaig regeneration; inline image, Dufftown; inline image, Dundonnell; inline image, Glensaugh; inline image, Otterburn; inline image, Redesdale – Burnhead mature; inline image, Redesdale – Burnhead regeneration 1; inline image, Redesdale – Burnhead regeneration 2; inline image, Redesdale – Road cut. inline image, inline image, inline image, initial heather cover ≥ 50%; inline image, inline image, inline image, inline image, 50% > cover > 35%; inline image, inline image, inline image, cover < 35%.

Download figure to PowerPoint

There was a similarly, although slightly less clear relationship between the rate of change in the proportion of heather and stocking rate (Wald/d.f. = 46·34, P < 0·001, Fig. 1b; Supporting Information, Appendix S2). This explained 66·8% of the variance at the site level and 5·1% at the within site level. Additional terms tried in the model – stocking rate2, summer/winter, region and phase – were not significant at P < 0·05 and did not reduce the deviance significantly. The derived relationship predicted that there would be no change in heather cover at 1·82 sheep ha−1 year−1. The 95% confidence intervals for the fitted line cross zero change at 1·14 and 2·61 sheep ha−1 year−1, indicating a substantial spread within the data.

As expected, the relationship between stocking rate and percentage utilization was strong (Wald/d.f. = 185·5, P < 0·001, Fig. 1c; Supporting Information, Appendix S2). The fitted relationship explained 87·2% of the variance at the site level and 78·2% at the within-site level. Utilization increased by 15·1% for each increase in sheep stocking rate of 1 sheep ha−1 year−1.

Discussion

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

heather dynamics and grazing

Both the utilization and stocking rate models were significant and fitted the data moderately well. However, as the majority of the experiments were established on communities with a high proportion of grasses and sedges (up to 50% in some cases), the estimated stocking rate for no effect, 1·82 sheep ha−1 year−1, is far higher than could be withstood by a moorland with a more complete cover of heather (Grant et al. 1982; Hartley & Mitchell 2005). The analysis based on the utilization of heather reveals a level of off-take that is directly related to heather performance, whereas to do the same for stocking rate would need further experiments that covered a full range of heather proportions in the vegetation as well as the full range of stocking densities. It may even have to include different associated grass species as well. Thus, developing models based on utilization data is more efficient than developing models from stocking rate due to the need to accommodate different vegetation types into the design of studies based on the latter.

management implications

This analysis, for the first time, synthesized data from a wide range of experiments where the impact of grazing on heather moorland had been assessed in a way that is both quantitative and relevant for the ongoing performance of the plant. The results are consistent with the interpretation of ‘heavy browsing’ of heather as being unsustainable in the longer term (MacDonald et al. 1998). The fitted relationship indicates that at 31·6% utilization, there should be no change in the proportion of heather in the vegetation. This value of 31·6% is substantially below the 40% utilization level that could be sustained in the short term at the Glensaugh site (Grant et al. 1982) and that was thought to be sustainable for more than moderately vigorous heather (equivalent to a Grazing Index of 66% of shoots browsed when average shoot growth > 4 cm year−1 on stands of younger regenerating heather, or where climate, soils and other environmental factors were more favourable to growth in general; Armstrong & MacDonald 1992; MacDonald et al. 1998). It is, however, above the 22% utilization derived from physiological models (Read et al. 2002) and the 20% utilization thought to be sustainable by less than moderately vigorously growing heather (< 4 cm year−1). It is, however, very similar to the estimate of 30·1% (Palmer et al. 2004) derived from analysis of two of these experiments (Glensaugh and Dufftown) and a long-term study of the impacts of uncontrolled grazing on moorland (Welch & Scott 1995), the results of which had to be converted to utilization using a pre-existing calibration (Palmer & Hester 2000).

The spread of points and the width of the 95% confidence intervals in Fig. 1, however, suggest that setting a ‘sustainable’ grazing level of 30% utilization could be too high. If, for instance, the spread of points was a consequence of different sites at equilibrium between heather and grass being able to support different amounts of heather utilization [c.f. the impact of browsing on willow Salix spp. is dependent on site conditions (Singer et al. 1994)], then there is a substantial probability that any individual site would show changes in the ratio of heather to grass at 30% heather utilization. As nitrogen has been shown to be a driver in heathland dynamics (Bobbink et al. 1998), it is possible that the spread of sites within the data was related to nitrogen deposition. However, all the sites used in this analysis were in areas with total nitrogen deposition rates below or just above (< 0·5 keq ha−1 year−1) the exceedance threshold (Hall et al. 2004), with no evidence that those just above the exceedance threshold (the Redesdale sites) behaved differently from the other sites (Fig. 1). Interaction between nitrogen and grazing may play a larger role in more southerly, more polluted moorlands.

Setting a site level utilization figure to where the lower confidence interval crosses the zero change line would be conservative, but would be appropriate where minimizing the risk of losing heather cover was important (individual managers might set different utilization figures if they had other goals than minimizing the risk of losing heather cover). Effectively, it assumes that a site manager with no further information suggesting that the site is more or less sensitive than average to grazing, would minimize any risks of an increase in the cover of monocotyledonous plants. This conservative approach reflects the fact that a shift towards monocots is more problematic in most circumstances than invasion by trees and shrubs, which rarely invade unless grazing is reduced to negligible amounts (Miles, Welch & Chapman 1978; Hobbs & Gimingham 1987). Consequently, setting a utilization level of 22·5% would, under most circumstances, appear to ensure sustainable use of the moorland resource in the long-term from a ‘nature conservationist’ or ‘wildlife management’ perspective (Mysterud 2006). It should be noted that the width of the confidence interval is determined by the sample size and that if more than ten experiments were available, then this conservative utilization figure might be higher.

Practical application of this conservative offtake level would equate to 20%, a halving of the first estimates made for sustainable grazing levels based on a single short-term experiment (Grant et al. 1978, 1982) and levels set for more than moderately vigorous heather (MacDonald et al. 1998). Setting the sustainable utilization lower than the predicted 30% no effect point is also essential as foraging by large herbivores is not spatially uniform, being concentrated on dwarf shrubs immediately surrounding grass patches (Hester & Baillie 1998; Palmer & Hester 2000). Also, the non-selection of heather age in the model building suggests that assuming a higher level of sustainable utilization for younger or more vigorous heather might be misplaced.

Analysing the response of vegetation to grazing based on the behaviour of grazing-intolerant species offers an efficient and powerful means of setting sustainable grazing regimes that could be applied in many different situations, particularly to those which are based on the biology of the competing species in the vegetation. Monitoring utilization and the consequent adjustment of stocking levels may be time-consuming compared to maintaining a set stocking levels, but it does offer the opportunity to be more precise in managing vegetation that is sensitive to inappropriate grazing regimes. This approach can integrate many processes that have to be taken into account when developing appropriate stocking densities, such as the proportion of intolerant vs. tolerant species in the vegetation, the relative preference of livestock for each and potentially how they are distributed in the landscape, and hence, avoids the need for experimental or survey-based data collection to assess different grazing regimes. It also offers a method of assessing the impact of wild herbivores on vegetation through comparison with the sustainable grazing offtake of grazing-intolerant species by domestic herbivores.

Acknowledgements

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

We thank Lynne Torvell for curating the records from many of the sites and Rob Brooker for commenting on an earlier draft. Comments by two anonymous referees, Jan Lepš and the editor greatly improved the relevance of the work presented. This work was funded by the Scottish Government's Rural and Environment Research and Analysis Directorate.

References

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

Supporting Information

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

Appendix S1. Ten grazing experiments on heather moorland spanning three regions of the UK

Appendix S2. Parameter estimates from the models shown in Fig. 1

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

FilenameFormatSizeDescription
JPE_1603_sm_AppendixS1-2.doc39KSupporting 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.