General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: research.competitions@defra.gsi.gov.uk

SID 5 Research Project Final Report

SID 5 (Rev. 3/06) Page 1 of 29

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code BD1228

2. Project title

Determining Environmentally Sustainable and Economically Viable Grazing Systems for the Restoration and Maintenance of Heather Moorland in England and Wales

3. Contractororganisation(s)

ADAS UK LtdNERC Centre for Ecology & HydrologyInstitute for Grassland & Environmental ResearchNewcastle UniversityRoyal Society for the Protection of BirdsScottish Agricultural College     

54. Total Defra project costs £ 1,044,830(agreed fixed price)

5. Project: start date................ 01 January 2002

end date................. 31 March 2007

SID 5 (Rev. 3/06) Page 2 of 29

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.1. Heather moorland (upland heath and mire) habitats are of international conservation significance and

have high biodiversity value, including characteristic upland bird and invertebrate assemblages. However, these habitats have declined in extent and condition due to heavy grazing and other inappropriate management. Livestock farming is a traditional and important business enterprise in the hills and uplands although its profitability has declined substantially during the last decade. In addition, a combination of agricultural support payments and market prices resulted in a significant reduction in cattle numbers in the hills and uplands in favour of sheep. The challenge is therefore to identify approaches, which may include a range of sheep, cattle and mixed grazing regimes, for restoring moorland habitats to the benefit of both biodiversity and economically viable hill farming. Until recently, agri-environment schemes have focused primarily on the reduction of livestock numbers, which has tended to halt the decline in dwarf shrubs such as heather (Calluna vulgaris), but has been less successful in reversing the process. This has been exacerbated by most grazing systems relying solely on sheep. To address this, farmers could claim enhanced payments under the Hill Farm Allowance Scheme for keeping cattle. If cattle numbers were to rise, they could have important impacts on upland vegetation communities, including the control of competitive grasses such as mat- grass (Nardus stricta) and purple moor-grass (Molinia caerulea) each of which can prevent the recolonisation of heather (Calluna vulgaris) and other dwarf shrubs. The overall aim of this project was to establish environmentally and economically sustainable grazing regimes that enable the restoration and maintenance of heather moorland habitats such as upland heaths and mires, and to produce practical guidelines to assist project officers and land managers in their implementation.

2. Current practices for moorland restoration include reduction in sheep numbers to arrest deterioration and introduction of summer cattle grazing in some situations. A scoping study (literature review and stakeholder consultation) was carried out to assess current knowledge and identify the key moorland habitats, the biodiversity, economic and recreational issues for degraded habitats and criteria for assessing the outcome of restoration management. Current practices were reviewed from eleven case study sites with wide geographic coverage in the Severely Disadvantaged Area in England and Wales.

3. There was broad agreement between model outputs and empirical studies. A modelling programme was carried out to explore the potential impacts of grazing regimes on biodiversity, livestock production and farm economics. Models were (i) a spatially explicit grid-based vegetation model, (ii) statistical models of bird species - habitat associations, (iii) ordination models of invertebrate group – vegetation associations and (iv) empirical models of livestock production and upland farm economics.

4. Vegetation modelling indicated no universal set of prescriptions for maintaining or reversing

SID 5 (Rev. 3/06) Page 3 of 29

the loss of heather moorland habitats. Fifteen scenarios were explored for two Calluna-Molinia and two Nardus-dominated sites at Redesdale (Northumberland), Molland Moor (Exmoor), Birkbeck Common (Cumbria) and Pwllpeiran (mid-Wales. These included three regimes with sheep only grazing (2.1, 1.5 or 0.66 ewes ha-1), two with summer only sheep (2.1 or 0.66 ewes ha -1 from May to August), two with summer only cattle (0.75 or 0.225 cows ha-1 in July and August), three with all year sheep and summer cattle (0.66 ewes ha-1 each with 0.75 or 0.225 cows ha-1 during summer and 1.5 ewes ha-1 with 0.75 cows ha-1 in summer), four rotational regimes with all year sheep (2.1 ewes ha -1) or summer only cattle (0.75 cows ha-1 in July and August) grazed for one or two consecutive years in five, and no grazing. Comparison of model outputs with field results revealed good agreement between the two sets. Scenarios were then run over a 20 year period to assess long-term vegetation response. The effectiveness of different livestock species in delivering vegetation management objectives was predicted to vary with vegetation type and spatial pattern. Summer cattle (0.75 cows ha-1) were effective in controlling Molinia in areas where it was dominant and in opening up the sward on Agrostis-Nardus heath to enable dwarf shrubs and other species to establish. However, if the latter were not present in the vegetation understorey or seedbank then Nardus may be expected to persist or increase in extent. Under low all-year sheep grazing (0.66 sheep ha -1), broad-leaved grasses (Agrostis sp.) were predicted to increase in extent and to out-compete Nardus. Reducing or removing sheep was effective in reducing Nardus extent, but was predicted to have little effect on Molinia. In each case the rate of vegetation change was influenced by the spatial distribution of the target species. For example, Molinia declined faster under summer cattle at Molland Moor, where it was more evenly distributed between plant communities, than at Redesdale, where its extent varied significantly between plant communities. Thus the rate and direction of change is dependent on the spatial distribution, composition and condition of the starting vegetation and recommendations on the use of particular grazing regimes need to be tailored to take these into account.

5. Bird-habitat models, produced across four upland regions (South Scotland, North Pennines, South Pennines and Wales), demonstrated the importance of heterogeneity in both vegetation composition and structure for promoting the abundance of the widest range of the nine species studied. Meadow pipit and whinchat abundance peaked at c.30 % heather cover whilst stonechat and red grouse abundance levelled off at c.40 - 50 %. Skylarks were most abundant where grasses dominated. Red grouse, whinchat and stonechat were associated with dense vegetation; whinchat, stonechat, curlew, meadow pipit and wheatear with heterogeneous structure; and golden plover, snipe, and skylark with open or short vegetation. Vegetation characteristic of wet flushes was associated with the abundance of both curlew and snipe. The direction and shape of these bird-habitat relationships were consistent across the four regions in most (c.90 %) cases, suggesting wide applicability, although models combining these terms had limited ability to predict species abundance when applied to new regions. Therefore, these findings can be used to guide grazing management to benefit moorland birds across UK uplands, but reliable, quantitative, predictions of abundance will generally be restricted to the four regions where data were collected.

6. Invertebrate modelling predicted that introduction of cattle grazing would cause the largest changes in relative abundance in many groups but various grazing regimes will be needed to maintain diverse invertebrate communities. Invertebrates comprise a significant component of moorland biodiversity and respond relatively quickly to habitat changes. They are also important in the diet of upland birds. Spiders, beetles, true-flies, bugs, Hymenoptera (sawflies, wasps, bees, ants), caterpillars and earthworms are widely taken, with true-flies and beetles being the most important (especially craneflies, ground beetles, weevils and click beetles). Invertebrate and plant field data from Redesdale (see para 8) were used to construct the invertebrate model. Annual summaries of plant species composition from vegetation model outputs over 50-year runs were then input to the invertebrate model. Four different grazing scenarios tested were sheep only (0.66 or 1.5 ewes ha -1) or mixed sheep and cattle (as sheep only plus 0.75 cows ha-1 in July and August). Separate model runs were created for spiders, phytophagous bugs, ground beetles, all beetles, true flies, true bugs and crane flies. The predicted changes after introduction of cattle reflected changes in vegetation. Mixed grazing with higher sheep stocking rates was predicted to cause a rapid relative decline in spiders (except in mixed heath). Phytophagous plant bugs respond to plant species composition and increased with this grazing scenario in Molinia-dominated vegetation, as did ground beetles. Relative abundance of true flies was less sensitive to grazing regime, although response of crane flies varied among plant communities.

7. Economics modelling showed that upland grazing systems will only be profitable with financial support. Existing bio-economic models for hill sheep systems were further developed to include cattle. Two elements were an energy based model working at the flock/herd level and a farm-scale model which fitted outputs into a simple farm context, but focussing on the moorland element. The same grazing scenarios were modelled as in the vegetation model. Different scenarios had highly divergent impacts on livestock numbers, labour demand and economic outcomes at farm, local and regional levels. Net margins without any supportive payments were shown to be deeply negative, especially the higher all-year sheep only scenarios (2.1 or 1.5 ewes ha-1) and high summer cattle scenarios (0.75 cows ha-1) with or without sheep. Addition of support payments moved some

SID 5 (Rev. 3/06) Page 4 of 29

scenarios theoretically into the positive but largely by reductions in livestock numbers and labour costs. With agri-environment payments, scenarios with cattle or low sheep stocking densities (0.66 ewes ha-1, summer only and rotational scenarios) moved into significant surplus, with high presumed levels of agri-environment support, but the sustainability of systems that require cattle for short summer grazing periods remains questionable.

8. System scale studies on Nardus grassland showed only small effects of grazing regime on vegetation and livestock performance. On Molinia heathland, mixed cattle and sheep grazing created suitable conditions for heathland restoration. Grazing regimes were tested in system-scale studies carried out in paddocks at Pwllpeiran (Nardus grassland) and Redesdale (Molinia wet heath) from 2003-06. Treatments at Pwllpeiran were low sheep (Welsh Mountain; 1.0 sheep ha -1), high sheep (1.5 sheep ha-1), cattle only (yearling Welsh Black heifers; 0.5 cattle ha-1 for two months in summer) and low sheep plus cattle (1.0 sheep + 0.5 cattle ha -1). Treatments at Redesdale were low sheep (Scottish Blackface; 0.66 ewes ha-1), low sheep plus cows (continental cross dry cows in summer; 0.75 cows ha-1), high sheep (1.5 ewes ha-1) and high sheep plus cows. Stocking rates tended to have more effect on individual animal performance than livestock species. Performance tended to be better at lower stocking rates. The response of Nardus grassland to different grazing regimes and the consequent effect on livestock performance will take several years to manifest. On Molinia heathland gains in cattle liveweight over the summer were sufficient for non-lactating cows, although the mixed regimes would not be sustainable at these stocking densities on an annual basis. Cows successfully reduced Molinia cover but wet heath vegetation was not re-establishing after four years. Molinia increased at the expense of Calluna in sheep-only paddocks. Invertebrate species composition was dependent on plant species composition and structure, which was partly attributable to cattle grazing.

9. Calluna (up to 15% ground cover) was successfully re-established in restoration experiments on Nardus and Molinia dominated moorlands within 3 years. Replicated plot-scale experiments within the system study paddocks at Pwllpeiran and Redesdale during 2003-2006 showed (1) the creation of bare ground (<10 % ground cover) by rotavation or trampling was essential for Calluna establishment; (2) where sites lacked a viable Calluna seedbank seed addition was essential (0.8 g m-

2 in this study); (3) seed addition improved the rate of Calluna establishment even when a natural seedbank was present; (4) soil disturbance resulted in a small increase in the presence of soft rush (Juncus effusus); (5) the best establishment and growth of Calluna was in the cattle only grazing treatment and ungrazed plots; (6) Calluna growth and seedling survival was reduced if sheep were present; (7) in Molina dominated grassland, removal of grazing allowed Calluna establishment but did not control grass regrowth.

10. Grazing studies confirmed the potential to exploit differences in dietary preferences, but highlighted the poor feeding value of heathland plants. Studies with two breeds of sheep (Welsh Mountain, Scottish Blackface) and two breeds of cattle (Welsh Black, Continental cross) as in the system studies were conducted at sites with Low (8%), Medium (36%) and High (61%) Calluna cover. The objective was to investigate interactions between animal type, sward composition and diet selection in order to improve understanding of factors influencing the impact of grazing on heather moorland, and to investigate associated effects on herbage intake, nutrient supply and rumen fermentation characteristics. Cattle and sheep were confirmed to be selective feeders, consuming grasses in preference to dwarf shrub. The quantities of Calluna consumed were small even on the High site, reflecting the availability of preferred food items. There were between-species differences in the contribution of key plant groups to the diet. There were also significant differences in the composition of the diets selected by the different breeds of sheep, whereas the diets of the two cattle breeds were generally similar. Blood metabolite profiling indicated that the intake of both energy and protein for all experimental animals was less than optimal to supply even maintenance requirements.

11. Conclusions:(i) Cattle can have a role in moorland grazing management.(ii) Cattle can reduce competitive grasses but are likely to damage sensitive habitats such as

blanket bog.(iii) Sheep at low stocking densities are better for vegetation management and individual animal

performance than at high stocking densities.(iv) Heterogeneity in moorland vegetation is required to maintain species diversity of birds and

invertebrates.(v) Outcomes depend on the grazing regime and vegetation characteristics so site-specific

regimes are necessary to meet particular objectives.(vi) Vegetation can take several years to show a substantial response to grazing.(vii) Rotational grazing regimes show promise for vegetation management.(viii) Restoration of heather requires disturbance and a seed source.(ix) Upland grazing systems, including those with cattle, require considerable financial support, to

achieve positive margins.(x) Summer only grazing systems, need a whole-farm approach to accommodate livestock all

year.

SID 5 (Rev. 3/06) Page 5 of 29

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

SID 5 (Rev. 3/06) Page 6 of 29

Determining Environmentally Sustainable and Economically Viable Grazing Systems for the Restoration and Maintenance of Heather Moorland in England and WalesReference Number BD1228.

C.N.R. Critchley1, O.D. Davies1, H.F. Adamson1, P.A. Anderson2, G.M. Buchanan3, M.D. Fraser4, S.M. Gardner1, M.C. Grant3, B.M.L. McLean1, R.J. Mitchell5, J.W. Pearce-Higgins3, R.J. Rose5, R.A. Sanderson6 & A. Waterhouse7.1ADAS UK Ltd, 2Penny Anderson Associates, 3RSPB, 4IGER, 5CEH, 6Newcastle University, 7SAC.

1. Introduction

Upland heath and mire habitats (generically referred to as heather moorland) in the UK are recognised as being of international conservation significance (Thompson et al., 1995) but have declined in extent and condition as a result of heavy grazing and poor management by burning or other practices (Bardgett et al, 1995). Only 14% of moorland in England is in favourable condition (English Nature, 2001), many becoming grass-dominated due to heavy grazing (e.g. Anderson & Yalden, 1981). Targets for the enhancement and maintenance of these habitats have been established within the UK Biodiversity Action Plan (BAP) and sites have been designated as Special Areas of Conservation (SAC). Heather moorland in the UK holds a unique assemblage of bird species, including eight species listed in Annex 1 of the EC Birds Directive 79/409/EEC. Recent evidence suggests that the UK populations of several bird species associated with moorland have declined in recent decades, with habitat change as a result of increased grazing pressure on moorlands considered a potential threat (Fuller & Gough, 1999; Sim et al., 2005). Invertebrates also form an important component of upland ecosystems, and can occur in significant numbers of both species and individuals. Many species are known to comprise an important part of the diet of higher animals, especially birds. Others have an impact on the vegetation; for example heather beetle (Lochmaea suturalis) can cause extensive damage to stands of Calluna vulgaris.

The principal challenge is to identify approaches for regenerating and restoring moorland habitats that are reliable, cost-effective and compatible with economically viable hill farming systems. Approaches used to date, implemented mainly via agri-environment schemes, have focused primarily on the reduction of livestock numbers, particularly sheep. These have been effective in halting the decline in cover of dwarf shrubs such as Calluna, but have had less success in reversing the process.Livestock farming is a traditional and important business enterprise in the hills and uplands. Since 1995, the profitability of farming in Less Favoured Areas (LFAs) has declined sharply, with cattle and sheep farms experiencing a 38% drop in net farm income between 98/99 and 99/00. Reduced stock numbers can increase individual sheep performance but this needs agri-environment scheme payments to compensate for reduced lamb output per hectare (Hyslop & Merrell, 2002). As current schemes are voluntary and sit alongside market and other policy based income, environmental benefits will need to be delivered by practical methods that minimise economic impacts.

As outlined in the Rural Development Plans for England & Wales, the government is committed to encouraging and supporting economically viable and environmentally sustainable businesses within the rural community, particularly LFAs. Thus, there is a need to identify environmentally sustainable and economically viable grazing systems in order to meet both the conservation and business enterprise requirements of these upland areas. This project set out to deliver: A sound scientific basis to assist Defra in developing appropriate policies and schemes to deliver UK BAP

targets for the enhancement and maintenance of upland habitats; Practical and credible guidelines to assist project officers and land managers in the implementation of grazing

systems and management strategies for the enhancement and maintenance of upland habitats.

1.1 ObjectivesThe overall aim was to establish environmentally and economically sustainable grazing regimes that enable

the restoration and maintenance of heather moorland habitats such as upland heaths and mires. Specific objectives were to:1. Complete a scoping study to include consultation exercises (to farmers, land managers, relevant associations,

government agencies), critical reviews of relevant scientific literature and current research/monitoring data.2. Evaluate existing grazing operations and intervention techniques for maintaining and restoring heather

moorland habitats, identifying the factors influencing their success, and determine their economic viability and impact upon aspects of biodiversity.

3. Develop, through modelling, single and mixed grazing regimes in conjunction with restoration techniques for restoring and maintaining heather moorland habitats in different geographical locations, and assess impacts of resulting habitat changes on invertebrate and bird populations.

4. Assess, through system scale and plot scale studies, the impact of selected grazing regimes and intervention techniques on vegetation, livestock and economic viability, upland birds and invertebrate species/groups of conservation importance.

5. Develop practical guidelines for both farmer/landowners and advisers on environmentally sustainable grazing regimes and report to Defra on the success of the project, and where necessary identify changes to scheme prescriptions and make recommendations.

SID 5 (Rev. 3/06) Page 7 of 29

Appendices to this summary report contain a full account of the project (including bibliography).

2. Scoping Study (Objective 1)

2.1 Literature ReviewA desk study was undertaken to collate current literature on moorland biodiversity and economics in relation to

livestock grazing systems. Specific topics reviewed were (i) grazing impacts on vegetation including the manipulation of grazing regimes for restoration, (ii) invertebrates of moorland in relation to habitat and management regimes, (iii) the diet of invertebrate feeding moorland birds, (iv) grazing behaviour of domestic and semi-domestic livestock in relation to dietary selection and digestibility, (v) bio-economic modelling including the linkage between biological and economic factors and (vi) the potential impact of policy change including CAP reform and cross-compliance measures. The literature review is in Appendix 1a.

2.2 WorkshopA workshop (invited participants) with relevant conservation (e.g. EN, SNH, CCW, RSPB, Biodiversity Steering

Group) and land management bodies (e.g. NFU, FUW, NSA, NBA, Moorland Association, RDS) was conducted on 6 June 2002. Workshop sessions were held on (i) definition and regional variation in degraded habitats, (ii) biodiversity requirements for degraded and restored habitats, (iii) economic requirements of degraded habitats, (iv) current restoration activities and their drawbacks and problems and (v) identification of a range criteria to assess the success of moorland restoration.

Outputs from the workshop were a list of habitats to be addressed in the study, with their biodiversity and economic requirements and a set of criteria to be used for assessing the success of existing grazing and intervention techniques in restoring the vegetation condition, biodiversity interest and economic viability of moorlands from agricultural and recreational activities.

Issues were raised concerning the definition of favourable condition for non-designated sites and the starting point and desired directional change for restoration. This was related partly to variable starting conditions resulting from historical changes and to the level of current knowledge. The need to define the end point(s) when a restoration regime switches to maintenance were discussed and it was concluded that this could only be done at present by professional judgement based on experience. Limited discussion about practical guidelines (the project output) included views on their content such as vegetation structure for different objectives, livestock breeds and economic viability. The detailed workshop report is in Appendix 1b.

3. Evaluation of Current Practices (Objective 2)

Stakeholder consultation helped to identify sites providing examples of various grazing regimes and heather restoration techniques that are currently being applied. From a list of potential sites, a subset demonstrating different grazing and intervention techniques were visited and assessed for habitat condition, landscape and shooting/recreation. Data were also collated, if available, on farm business and livestock performance. Assessments used criteria identified in the workshop, along with the standard upland habitat condition assessment method (JNCC, 2004). The resulting eleven farm case studies were prepared to provide benchmarks for the development of regimes for restoring and maintaining heather moorland (Appendix 2). The information gathered during the consultation exercise and from farm visits and assessments were collated and used to identify a range of grazing scenarios that might be used to maintain and restore heather moorland habitats. The potential effects of these grazing scenarios were then modelled at four of the case study sites (Objective 3).

A wide range of habitats was assessed, including dry dwarf shrub heath, wet heath, blanket bog and valley bog. Sites were located in Northumberland, North Pennines, Lake District, Forest of Bowland, North York Moors, Peak District, mid-Wales, Exmoor and Dartmoor. Most sites had been degraded in the past by heavy grazing and sometimes by inappropriate burning, but with grazing regimes now adjusted to aid habitat restoration. Reduced sheep stocking levels, especially in winter, tended to arrest deterioration of dry heath, wet heath and blanket bog but there was only limited evidence of re-establishment of dwarf shrubs and other characteristic plant species. Very low stocking levels resulted in an increase in biomass, particularly of grasses. Introduction of summer grazing by cattle was done at several sites on both dry heath and wet heath, and tended to reduce grass biomass, to the potential benefit of dwarf shrubs. Reduction of previously high cattle stocking levels in winter was also beneficial. However, in most cases the adjusted grazing regimes had only been implemented for a few years and there had been insufficient time for major habitat changes to take place.

Sites used for modelling included the two ADAS Research Centres (ADAS Pwllpeiran in the Cambrian Mountains and ADAS Redesdale in Northumberland), which were also used for the field experiments (Objective 4). In addition, Molland Moor in Exmoor (Case Study 10, Appendix 2) and Birkbeck Common in the Lake District were used to model grazing scenarios. Additional data on vegetation, stock numbers and type, and animal performance were collected from these sites.

SID 5 (Rev. 3/06) Page 8 of 29

4. Refine Models, Develop Grazing Regimes and Assess Impacts via Modelling (Objective 3)

The aim of the modelling programme was to explore the potential impacts of grazing regimes on biodiversity, livestock production and farm economics and assess their potential to deliver benefits in different upland vegetation types. The modelling programme included four elements:1. A spatially explicit grid-based vegetation model examining the interaction of grazing, plant competition and

species distribution on upland vegetation dynamics;2. Separate statistical models establishing the relationships between individual moorland bird species,

vegetation, management & environmental variables;3. Multivariate ordination models determining the associations between vegetation composition and different

invertebrate groups;4. Empirically based models of livestock production and upland farm economics.

The models were operated separately but the aim was also to link them via their inputs and outputs (Fig. 1). Baseline field data were used to initialise the vegetation model and for prediction of long-term effects. Outputs from this model were then used to initialise the bird, invertebrate and production models. Outputs from all three modelling streams were used along with other project outputs to evaluate the potential sustainability of different moorland management scenarios for farming and biodiversity.

Fig. 1. Links between the different modelling elements.

4.1 Vegetation ModellingA hierarchical grid-based modelling approach was used to explore the dynamics of upland vegetation change

under different grazing regimes. The basic unit of the grid is the cell, each of which is occupied by a single plant species of known age. Cells are grouped into tiles representing different plant community types and these are then distributed across the spatial grid in a manner simulating the pattern of plant communities present in the field. Within the model, plant species change is driven by livestock grazing, plant competition and the spatial distribution, age and starting abundance (number of cells occupied) of each plant species present. The vegetation model developed under earlier Defra contracts (BD1211 and BD1218) was designed to assess the effect of grazing by single animal species alone (either sheep or cattle; Gardner, 2002). This model was re-written to enable the spatial grid to be grazed simultaneously by sheep and cattle. Outputs were also refined to enable predictions of species change within individual plant communities as well as the whole site. Further details are given in Appendix 3a.

Selection of system study grazing regimesResults from preliminary model runs were used to select grazing regimes for the system studies at ADAS

Pwllpeiran and ADAS Redesdale (see Sections 5.1 and 5.2 below for site descriptions and management objectives). Spatial grids were prepared to simulate vegetation at the system study sites and a range of all-year sheep only, summer sheep or cattle only and mixed sheep plus cattle grazing scenarios at different stocking densities were assessed.

Comparison with field dataModel outputs and field data from Pwllpeiran and Redesdale were compared at both whole moor and plant

community level. Analyses focused on comparison of predicted and observed rates of species change rather than estimates of species abundance (percentage of cells occupied by each species), since field methods used for estimating species top cover and frequency were not directly comparable.

Overall predicted change in key species (Calluna, Molinia caerulea, Nardus stricta, Vaccinium myrtillus, Broad-leaved grasses and Rushes) at both sites fitted well with field data collected from the different systems study treatments. A single descriptive example for the high sheep plus cattle treatment at Redesdale is presented in Fig. 2. Further results are given in Appendix 3a.

SID 5 (Rev. 3/06) Page 9 of 29

Fig. 2. Field vs. model data for high sheep (1.5 ewes ha -1) plus cattle (0.75 ha-1 summer only) grazing regime at ADAS Redesdale for a) Calluna and b) Molinia. % abundance refers to top cover and frequency (field data) and cell occupancy (model data) at the whole moor scale.

Development of grazing scenariosFifteen grazing scenarios were chosen for modelling. These included:

1. Three all year sheep only regimes using a) 2.1 (High), b) 1.5 and c) 0.66 (Low) ewes ha -1 (each regime having a 25% reduction in sheep numbers November – March inclusive and accounting for short-term variation in sheep numbers due to lambing, etc.);

2. No grazing all year;3. Two summer sheep only regimes using a) 2.1 or b) 0.66 ewes ha-1 from May –August inclusive;4. Two summer cattle only regimes using a) 0.75 or b) 0.225 cows ha -1 in July and August;5. Three mixed regimes combining all year sheep regimes of a) 0.66 or b) 1.5 ewes ha-1 with two summer only

cattle regimes of a) 0.75 or b) 0.225 cows ha -1 in July and August. The combination of sheep b) and cattle b) was not investigated;

6. Four rotational grazing regimes with either a) all year sheep @ 2.1 ewes ha -1 or b) summer only cattle @ 0.75 cows ha -1 in July and August. Animals grazed a site for either i) one year or ii) two consecutive years in a five year cycle.

The full set of scenarios was run for both Pwllpeiran and Redesdale and a subset for Birkbeck Common (regimes 1b, 1c, 2, 4a, 4b, 5a)a) and 6bi) and Molland Moor (regimes 1b, 1c, 2, 4a, 4b, 5a)a)). A baseline scenario (reflecting current management practice) of 2.84 ewes, 0.06 ponies and 0.048 cows ha -1 and two further scenarios using 0.66 sheep and 0.06 ponies ha-1 or 0.75 cows and 0.06 ponies ha-1 were run at Birkbeck Common. The baseline scenario for Molland Moor involved annual grazing of 0.147 cows from May-October, 0.63 ewes from June-September, 0.022 ponies and 0.44 deer grazing all year. Each site differed in the vegetation communities present and in its management objectives.

Pwllpeiran – grazing scenariosAt Pwllpeiran, the vegetation management objectives were to reduce the abundance of Nardus, restore dwarf

shrubs and maintain the diversity of other plant species. Model predictions suggested that the rotational (6ai, 6aii, 6bi) and low (1c) sheep grazing scenarios would meet the objectives most effectively. Under these scenarios a steady increase in Broad-leaved grasses (8-9% over 20 years at the Whole Moor scale) and a slow but steady decline in Nardus (8-9% over 20 years) was predicted. Nardus also declined under summer cattle only or mixed or grazing scenarios (4a, 4b and 5a) but by a smaller percentage (3%). Broad-leaved grasses were predicted to

SID 5 (Rev. 3/06) Page 10 of 29

0 . 0

1 0 . 0

2 0 . 0

3 0 . 0

4 0 . 0

5 0 . 0

0 1 2 3 4 5

Y E A R

% a

bu

nd

an

ce

C a l l u n a - m o d e l C a l l u n a - f r e q C a l l u n a - t o p c o v e r

a )

0 . 0

1 0 . 0

2 0 . 0

3 0 . 0

4 0 . 0

5 0 . 0

0 1 2 3 4 5Y E A R

% a

bu

nd

an

ce

M o l i n i a m o d e l M o l i n i a - f r e q M o l i n i a - t o p c o v e rb )

show a small decline (1-3%) under scenarios 4a, 4b and 5a and Vaccinium to show a small increase (5%). In addition, the occupancy by rushes was predicted to increase by 7-10% under scenarios with cattle present.

Species change within the three Eriophorum vaginatum, Nardus or Carex nigra dominant communities was broadly similar to that predicted for the whole moor. Model outputs, using data from Year 20 of each scenario, suggested that competition might occur between Nardus and broad-leaved grasses in Nardus- and Carex-dominated communities. Predicted competitive relationships did not differ between sheep and cattle grazing regimes, but cattle tended to be associated with a high occupancy of Nardus and sheep with a high occupancy of broad-leaved grasses. Since both species tend to prefer broad-leaved grasses in July and August, (though cattle are less selective and forage on larger grass patches), the lower occupancy of broad-leaved grasses under cattle might be due to their larger energy requirement and hence greater offtake. By reducing the occupancy of competitive broad-leaved grasses, cattle (scenarios 1b and 1c) might enable Nardus to persist longer in the sward, particularly in the absence of competitive dwarf shrub species such as Calluna (cf. Birkbeck Common). Under 1.5 and 0.66 ewes ha-1, broad-leaved grasses were predicted to increase and Nardus to decline. Given the low preference of sheep for Nardus, this decline is more likely to be attributable to competition between Nardus and other grasses rather than to direct grazing of Nardus by sheep.

Redesdale – grazing scenariosThe management objective at Redesdale was to reduce the abundance of Molinia to facilitate the restoration

of dwarf shrubs and the mosaic of wet heath plant communities. At Redesdale, summer cattle and mixed grazing scenarios 4a, 5aa & 5ab, were predicted to be the best for achieving these objectives, with Molinia occupancy at the Whole Moor scale predicted to be maintained or slightly reduced (2%) over a period of 20 years, and Calluna to increase steadily (6-7%). By contrast, under sheep only (1a-1c) and no grazing scenarios (2), Molinia was predicted to increase steadily (10 -14%) and Calluna to be maintained or to increase slightly (0-4%) over the same period. These differences were more evident at the community scale. Thus under cattle (4a, 5aa & 5ab), Molinia was predicted to decline markedly in the Molinia-dominated (6-7%) and mixed heath (6-9%) communities and to increase in Nardus, Juncus-, Calluna- and Carex-dominated communities (1-7%). Under sheep (1a-1c) and no grazing (2), Molinia was predicted to increase in all six communities by 4-22%. Calluna, under cattle (4a, 5aa&ab), was predicted to increase by 1-14% in all six communities and under sheep (1a & 1b only) to decline by 1-3% in the Calluna-dominated community, to increase by 1-3% in Molinia-, Nardus-, Juncus- and Carex- dominated communities and to remain almost constant in the mixed heath community (1a only).

Data for the baseline starting abundance for Calluna and Molinia provided some evidence for direct competition between the two species where one or both were abundant (Fig. 3). No competitive relationship was evident in Nardus-, Juncus- and Carex-dominated communities, where abundance of both species was low.

Fig. 3. Predicted end-points for Calluna & Molinia abundance in three plant communities after 20 years of different sheep (•) and cattle (•) grazing regimes with baseline starting positions (•) in each community.

Under the no grazing scenario, the gradient of the competitive relationship was predicted to increase, suggesting that Molinia would replace Calluna at a faster rate under this regime than is currently the case for the baseline data. A large predicted increase in the intercept value suggested that Molinia might also be competing with and replacing other species such as Nardus, Carex and Mosses. Thus, no grazing might benefit Molinia by increasing its competitive advantage over other species.

Comparison of predicted Calluna and Molinia abundance under different sheep and cattle regimes, suggested that Molinia occupancy, in communities where it is abundant, would show a greater increase under sheep grazing than under cattle (P<0.001; Fig. 3). Under scenarios with high summer cattle (4a, 5aa and 5ba), the competitive relationship was almost unchanged from the baseline position suggesting that high cattle grazing might favour Calluna by suppressing the increase in Molinia occupancy. Under low summer cattle or rotational scenarios (4b, 5ab, 6bi and 6bii), the relationship did not differ markedly from that predicted for no grazing (2).

Intercept values for sheep only (1a-1c, 3a & 3b) and low or rotational cattle scenarios (4b, 5ab, 6bi and 6bii) were similar to no grazing (2), suggesting that the increase in Molinia is also driven by its replacement of other

SID 5 (Rev. 3/06) Page 11 of 29

plant species. This was not predicted for high summer cattle scenarios (4a, 5aa and 5ba). In Nardus-, Juncus- and Carex-dominated communities where little Calluna-Molinia competition was predicted, both species appeared to replace species such as Carex, Deschampsia flexuosa, fine-leaved grasses and Nardus.

Birkbeck Common – grazing scenariosBirkbeck Common is situated within the Lake District Environmentally Sensitive Area (ESA) and covers an

area of 714 ha comprising rough acid grassland dominated by Nardus, with smaller areas of bent/fescue grassland, blanket bog, wet heath, valley bog, heather heath and rush-pasture and rank grassland. Most commoners have entered ESA management agreements. The management objective for the site is to reduce grazing pressure, restore the condition and possibly extent of dwarf shrubs (both Calluna and Erica tetralix) and reduce Nardus. Ponies were included in the grazing scenarios as these are part of the current management regime. Ponies were treated in the model as small dry cows (two-thirds of a dry cow applied to the site) because grazing preference and nutritional data were not available.

At Birkbeck Common, summer grazing by cattle (scenarios 4a, 4b, 5aa and 4a+ponies) was predicted to deliver a greater enhancement of dwarf shrubs (Calluna and Erica tetralix) than low sheep only scenarios (1b & 1c). However, the model also suggested that summer cattle grazing, particularly at a rate of 0.75 cows ha -1, could lead to significant increases in rush occupancy. With the exception of the bent/fescue grassland community, no apparent difference was predicted between low sheep (1b & 1c) and summer cattle (4a, 4b & 6bi) grazing on Nardus occupancy. Nardus was predicted to decline in areas where it was dominant and to show a small increase in areas where it was rare or absent under all single species grazing scenarios. A similar outcome was predicted for all mixed grazing regimes (5aa, 4a+ponies, 1c+ponies) but with a slightly smaller decline and larger increase where it was dominant or rare respectively.

As noted for the Pwllpeiran site, the model did predict a difference in the effect of sheep and cattle on Nardus for the bent/fescue grassland community. This was apparently due to the impact of cattle on broad-leaved grasses. Under cattle scenarios (4a, 4b & 6bi), the occupancy of broad-leaved grasses was predicted to decline by 31% and Calluna, rush and Nardus occupancies to increase. Of these, Nardus showed the smallest increase (4%). A similar effect was predicted for the blanket bog community. Here summer cattle were predicted to reduce Eriophorum occupancy by 18% and to increase the occupancy of Erica tetralix (4-14%), rushes (5-16%) and Calluna (6-8%). Nardus occupancy was also predicted to increase but only to an extent similar to that of low sheep scenarios (1b & 1c). Thus, by reducing the occupancy of broad-leaved grasses and Eriophorum in bent/fescue grassland and blanket bog, cattle on this site may provide an opportunity for the expansion of dwarf shrubs and rushes. This effect was not predicted under low sheep scenarios (1b & 1c). Under low sheep grazing, the occupancy of broad-leaved grasses was maintained or enhanced and where these are dominant, they may limit the expansion of Nardus.

Molland Moor – grazing scenariosMolland Moor covers an area of 681 ha of mainly Molinia-Calluna mixed heath vegetation. The site occurs

within the Exmoor National Park and is part of a privately owned estate. It is also within the South Exmoor SSSI and the Exmoor Heaths SAC and is under an Exmoor ESA agreement (Tier 1, Part 5), which places stocking restrictions on heather moorland. Molland Moor was one of four regional demonstration sites established by the Heather Trust (under Defra funding) to transfer knowledge on best practice management techniques and results from scientific research to end-users. Several areas of Molland Moor have been burned in the past, which has influenced the dominance of Molinia and Calluna present. The vegetation data used in this modelling study were based on habitat survey and detailed floristic assessments undertaken previously in seven areas of the moor (PAA, 2004). These areas included former and recent burned areas, mature Calluna, Molinia-dominated wet heath and mixed Calluna-Molinia moorland.

Molland is grazed mainly by ponies and red deer with sheep and cattle present in mid-late summer. Both deer and ponies were included in the baseline scenario but not in the standard set of sheep and cattle modelling scenarios. Both species were treated as small dry cows (two-thirds of a dry cow) in the model in terms of their daily energy requirements. With respect to grazing preferences, deer were considered to be more selective grazers like sheep and ponies to be larger ‘patch’ scale grazers like cows. This is an over-simplification of deer and pony grazing behaviour but reflects a lack of available quantitative data on their grazing preferences and nutritional requirements.

Predicted vegetation changes under the baseline scenario were a decline in dwarf shrub occupancy – particularly for Calluna (6%) and Vaccinium (13%), a significant increase in Molinia (24%) and a continued frequent occurrence of Deschampsia. Under low sheep scenarios (1b and to a lesser extent 1c), Calluna and Vaccinium were predicted to increase (6%). A small overall increase in Molinia (3%) was also predicted whilst Deschampsia and other fine-leaved grasses declined (5-8%). Under summer cattle and mixed scenarios (4a, 4b and 5ab), dwarf shrubs were predicted to show a greater increase in occupancy (13%) than under low sheep grazing whilst Molinia was predicted to decline significantly (14%). These trends were consistent across all plant communities, the response varying in magnitude between the seven communities.

4.2 Bird ModellingThe relationships between bird abundance and vegetation condition have been described recently for nine

relatively widespread moorland bird species, using data largely from south Scotland (Pearce-Higgins & Grant

SID 5 (Rev. 3/06) Page 12 of 29

2006). The work undertaken in BD1228 extended this approach by incorporating data from three further upland regions (North Pennines, South Pennines and Wales). Nine species were sufficiently common (present on > 20 % of plots) to allow analyses of abundance to be undertaken - red grouse, golden plover, curlew, snipe, meadow pipit, skylark, stonechat, whinchat and wheatear. Analyses were also conducted on the presence/absence of a further six species present on 10 - 20 % of plots, but these provide only relatively coarse assessments and are not considered further here (but see Appendix 3b).

The objectives of the bird-habitat studies were to: (i) assess the generality of relationships between moorland bird abundance and vegetation condition across regions; and (ii) produce bird-habitat models that can be used to predict likely impacts of different grazing systems on bird abundance, via linkage to modelling and systems studies that assess grazing impacts on vegetation.

The generality of individual relationships between bird abundance and vegetation condition was assessed following a two-stage analytical approach (Appendix 3b). First, models of bird abundance were produced using only non-vegetation variables (i.e. relating to topography, management, etc). Second, the significance of a range of vegetation variables when added to this stage-one model was used to describe the suite of vegetation relationships to which a particular species may respond. At this stage, the significance of a regional interaction with each of the vegetation variables was used to test the generality of the form of that relationship between the four regions. The bird-habitat relationships largely followed those identified by Pearce-Higgins & Grant (2006). In relation to heather cover, meadow pipit and whinchat abundance peaked at c.30 % cover whilst stonechat and red grouse abundance levelled of at c.40 - 50 % cover. Skylark were most common on non-heather (grass and sedge dominated) habitats. Vegetation structure was correlated with abundance for all species: red grouse, whinchat and stonechat were each associated with dense vegetation, although the densities of both whinchat and stonechat, and curlew, meadow pipit and wheatear were also enhanced by heterogeneity in sward structure. This indicates a wide benefit of providing some short, open vegetation within a tall sward, whilst golden plover, snipe, and skylark were each associated more specifically with such open or short vegetation.

The direction and form of these bird abundance-vegetation relationships were consistent across the four study regions in the majority (c.90%) of cases, indicating the generality of these findings to the UK uplands. Golden plover was the exception, with four of the 12 significant vegetation effects showing significant regional variation. However, in all four cases this related to vegetation cover, with vegetation structure remaining the most important determinant of golden plover abundance across all regions. Different vegetation types in different regions were responsible for the critical short, open vegetation required.

Whilst individual bird habitat relationships show wide generality, it was also necessary to test the predictive ability of models that considered the combined effects of variation in vegetation and non-vegetation variables. Such models were required to allow linkage with the modelling and systems studies to determine the effects of grazing systems on vegetation, and hence predict impacts on bird abundance. Previous work showed that such models often have good predictive ability within a region (Pearce-Higgins & Grant 2006), but it was important to assess the generality of models predicting overall bird abundance between regions. A model was built of predicted bird abundance using data from the two regions with the most data (South Scotland and the South Pennines), and the strength tested of the correlation between observed and predicted abundance in two independent test regions (North Pennines and Wales). The predictive abilities of two modelling approaches were compared at this stage; model averaging, which has recently become available as a technique for ecologists (Whittingham et al. 2005, 2006), and a more traditional backwards deletion approach (Appendix 3b).

Model predictions tended to show a closer fit to the observed data across the two independent test regions using model averaging (P = 0.009). Predicted abundance from the model-averaged models was significantly correlated with observed abundance for 9 of the 18 species-region combinations. For both red grouse and stonechat, the models performed well in both regions, whilst for snipe and meadow pipit; two of the species with greatest variability in measuring abundance (Buchanan et al. 2006, Hoodless et al. 2006), models had little predictive power. Where models had significant predictive power, the predictions explained an average of 46% of the deviance in bird abundance. Thus, on average, models are likely to have a 50% probability of being applicable to new regions (e.g. the North Yorkshire Moors), based upon the significant predictions above, but in some of these cases the predictive power may still be relatively low.

Based on these findings, final models of bird abundance were produced (using model averaging and the data from all four regions) to predict the impact of grazing regimes upon moorland birds. The original aim was to link these final models to the outputs from the vegetation dynamics modelling (Appendix 3a). However, to achieve this it was necessary to be able to predict how grazing systems affect vegetation structure as well as composition. Preliminary analysis of the vegetation data collected for the bird-habitat modelling suggested that vegetation structure could be predicted as a function of species composition and age, but attempts to model how grazing reduced vegetation height and density proved unsuccessful (Appendix 3a). Consequently, linkages have not been made between the vegetation dynamics and bird-habitat models.

4.3 Invertebrate ModellingOutputs from the vegetation model, in terms of annual summaries of the plant species composition, were used

as inputs into the invertebrate model. Invertebrate samples collected from pitfall traps at the Redesdale system study site were used to construct the model by relating the abundance of invertebrates to the different plant communities identified in the vegetation model. The four different grazing scenarios applied in the Redesdale system study were tested, over 50-year runs: low sheep (0.66 ewes ha-1), high sheep (1.5 ewes ha-1), low sheep

SID 5 (Rev. 3/06) Page 13 of 29

plus cows (0.66 ewes ha-1, 0.75 cows ha-1) and high sheep plus cows (1.5 ewes ha -1, 0.75 cows ha-1). As the species composition of the vegetation changed under the various grazing regimes each year, so these changes were directly translated into changes in both the species composition and abundance of the different invertebrates. Separate model runs were created for Araneae (spiders), Auchenorrhyncha (phytophagous bugs), Carabidae (ground beetles), Coleoptera (all beetles), Diptera (flies), Heteroptera (bugs) and Tipulidae (crane flies). Details of the model structure and results are given in Appendix 3c. Model runs show change compared to year zero in relative predicted abundance of each taxon (e.g. Fig. 4).

Fig. 4. Predicted change in relative abundance of Tipulids in 6 plant communities at Redesdale under a) high sheep plus cows and b) high sheep grazing regimes.

The largest changes in predicted relative abundance were generally after the introduction of cattle; this probably reflects the increased magnitude of shifts between the different vegetation communities with cattle grazing. The combination of cattle with high sheep stocking rates resulted in a decline in relative abundance of invertebrates in most plant communities. High mixed grazing was predicted to cause a rapid relative decline in Araneae (spiders) in all plant communities, except mixed heath, probably as a direct result of the changes in vegetation structure from this grazing regime. Phytophagous plant bugs (Auchenorrhyncha), which respond primarily to vegetation species composition, were predicted to show an increase with this grazing scenario in Molinia-dominated vegetation, with a more gradual decline on the mixed heath. Carabidae (ground beetles) showed a similar response, with largest changes in the high mixed grazing, whereas predicted relative abundance of Diptera as a whole (true flies) was less sensitive to grazing regime, being fairly similar across most treatments. Within this order, however, more subtle differences are apparent for the Tipulidae (Fig. 4), especially in the high mixed regime. The considerable impact of this grazing regime on the invertebrates as a whole probably reflects the greater community change within the vegetation model that was apparent for most plant communities.

4.4 Economics ModellingOver the duration of this study, the economic drivers and support mechanisms for hill and upland land

management changed profoundly. The introduction of the Single Farm Payment (SFP) as a de-coupled payment replacing direct subsidies was a major change. The ‘freedom to farm’ implicit within the reforms makes it feasible to consider a much wider range of livestock management systems and financial support mechanisms than previously. In tandem with the introduction of the SFP and continuation in a modified form of LFA payments, new agri-environmental schemes were developed within each of the countries of the United Kingdom.

The aim of this part of the study included the economic modelling of a series of scenarios based on the field experimental sites (Pwllpeiran and Redesdale) and case study sites (Birkbeck and Molland). Full details are in Appendix 3d; examples from Redesdale are presented here. The economic modelling required assumptions and significant planning with respect of the farming and labour systems and potential management criteria of different schemes for each of the scenario case studies in which they resided. The work used an existing computer based bio-economic model for hill sheep systems (Conington et al., 2004) to which new algorithms for cattle were added. The model (both existing sheep and new cattle) featured two core elements, firstly an energy based model focussing on hill pasture systems (which fed energy requirement data to the vegetation model), this working at the flock/herd level and then secondly a farm-scale model which fitted the outputs from this model into a simple farm model context. This latter scaled up the plot data to a partial farm, bearing in mind the elements of the system outside the ‘moor’ of study. Different scenarios were run, for which the main issue and the main set of assumptions dwelt upon this time off the ‘moor’. For both field experimental and case studies, livestock are removed from the moor for a range of management reasons including mating, lambing, grazing twins on inbye and removal of cattle for all but the main summer season. The focus of this work was to determine the economic considerations of the ‘moor’, not of the inbye area or the whole farm. This was because a) the moor was the focus of the study including the other modelling exercises and b) a wider and much more presumptive series of assumptions would have been needed to move to whole farm situation. The SFP, LFA and agri-environment schemes also pose considerable difficulties in terms of how they are included in economic scenarios. For

SID 5 (Rev. 3/06) Page 14 of 29

example, stocking rates under the English Higher Level Stewardship Scheme (HLS) are site-specific, with various supplements potentially available. For widest applicability and simplicity, rather than building layer upon layer of assumptions into the scenarios, a simplistic approach for eligibility was taken for scheme elements.

Modelling exercise 1There were five scenarios, using the Redesdale system study (72 ha, scaled up to 200 ha). These were

selected deliberately to be extreme in relation to evolving policy mechanisms and possible land manager strategies;1. High sheep stocking (aiming to maximise sheep production system without agri-environmental support, but

retaining SFP) – 2.1 ewes per ha, the historical level of grazing for the site.2. A mixed grazing model with 0.66 ewes ha-1 plus 0.75 cows per ha for 2 months (to retain SFP and obtain the

HLS payment of £40 ha-1) – coinciding with one of the grazing treatments at the site.3. Low summer only sheep (aiming to minimise sheep numbers, whilst retaining SFP) – sheep at 0.25 ewes

ha-1 4. An alternative lower intensity HLS strategy, mixing the ‘minimise sheep’ with added cattle - 0.25 ewes ha-1

plus 0.2 cows ha-1 for summer.5. Abandon the land with no income – no grazing

Fig. 5a. Income elements (no grazing has nil income), £ per 200 ha annum-1

Fig. 5b. 'Net Margin' (income less variable & labour costs), £ per 200ha annum-1

This modelling exercise predicted an annual financial turnover of £37,603, £53,508, £9,647, £19,987 and zero, for the five scenarios respectively. The first four scenarios included CAP area payments (SFP and LFA) of £7878, and second and fourth had payments of £8,000 from HLS (Fig. 5a). Using standard figures for labour requirements per livestock head (Stott et al., 2005) and the number of stock involved, estimates of farm labour requirements were made of 0.37, 0.54, 0.02, 0.14 and zero labour units/year to manage the stock on the hill. Margins were calculated for total income less variable costs and an estimate of labour costs only (Fig. 5b). As the moorland component of a farm is responsible for a low proportion of many farm fixed costs, this provided one way of comparing these systems. The ‘minimise sheep’ scenario produced better margins than the ‘maximise sheep’ system even if the two CAP payments were re-included. True decoupling, removing the CAP money (of £7878) to a separate account, would leave the maximise sheep scenario with a large deficit, unless the money were notionally re-coupled within the farm accounts.

Modelling exercise 2Grazing scenarios (see Section 4.1) were modelled for 4 sites using the assumptions and approaches tested,

and validated with stakeholders. In this exercise a proportion of other whole farm fixed costs (machinery, rents, buildings) were included to produce net margins, rather than only labour.

The results illustrate the dramatic and divergent impacts that different management scenarios will have on livestock numbers, labour demand and on economic outcomes at farm, local and regional levels. These exercises both confirm that net margins without recoupling of SFP and LFA payments and any agri-environmental payments will be deeply negative (Fig. 6a). Re-coupling the relevant SFP and LFA payment within the farm account moves some of systems theoretically into the positive, but largely by reductions in livestock numbers and subsequent costs (particularly labour). Unless this labour is truly variable (e.g. re-deployed or not hired in the first place) then the whole farm position will not improve, even though marginal economics of the moorland does. Agri-environment payments do have a major impact on margins as modelled (Fig. 6b). Some scenarios move into significant surplus, with low labour inputs and could easily withstand the true decoupling (and eventual disappearance) of SFP and LFA. By contrast with the traditional livestock management scenarios, the agri-environment schemes modelled here had lower inputs of labour, especially seasonally, and much lower outputs of livestock. Fundamental questions arise as to the sustainability of such systems through the time when the stock are not present on the moorland. The scenarios propose high numbers of cattle for short summer grazing periods and scaling these systems up would require significant numbers to be retained on non-moorland farmland and housing for the balance of the time.

SID 5 (Rev. 3/06) Page 15 of 29

Fig. 6. Net margins for Redesdale a) without SFP and HLS and b) re-coupling SFP and with possible HLS payments including higher cattle supplements.

5. Assess Impacts through System Scale and Plot Scale Studies (Objective 4)The overall aim was to field test selected regimes, identified under Objective 3 as likely to be successful in

restoring degraded moorland vegetation. Regimes were selected following preliminary runs of the vegetation model. Systems studies were conducted in fenced enclosures at ADAS Pwllpeiran in the Cambrian Mountains ESA, Wales, and ADAS Redesdale in the Northumberland National Park. The objective was to investigate the performance of cattle and sheep under different grazing regimes and assess implications for vegetation and other biodiversity. Restoration experiments were conducted on plots nested within the system studies at both sites.

5.1 Pwllpeiran System StudyThe site comprised 72 ha of Nardus grassland (similar to the Nardus stricta – Galium saxatile (U5) community;

Rodwell, 1992), formerly dwarf shrub heath degraded by past heavy grazing. The main objective was to reduce the abundance of Nardus, to restore dwarf shrubs such as Vaccinium and Calluna, and maintain the diversity of other plant species. A series of replicated plots was set up. The treatments were low sheep (1.0 sheep ha -1), high sheep (1.5 sheep ha-1), low sheep plus cattle (1.0 sheep + 0.5 cattle ha -1 for two months in summer) and cattle only grazing (0.5 cattle ha-1 for two months). Treatments were replicated over three blocks. Welsh Mountain ewes and yearling Welsh Black heifers were used. Cattle were turned out for grazing for two months during the summers of 2003 to 2006 inclusive. Cattle liveweights and condition scores were recorded pre and post summer grazing in each year. Lamb birthweights, shearing weights and weaning weights and ewe tupping and weaning weights and condition scores were also recorded. Data were analysed using REML (GENSTAT).

For vegetation assessments, a grid of 125 1 x 1 m quadrats at 75 m spacing was superimposed on the 12 plots. Presence and top cover of all plant species, litter and bare ground was recorded pre-treatment in 2002 and at the end of the experiment in 2006. Top cover was also recorded in 2004. Local shoot frequency (presence in each 10 cm cell) and a grazing index (% of occupied cells grazed) of four key species were recorded annually from 2003. All records were made in September-October. Treatment effects on the overall plant community were analysed by partial redundancy analysis with Monte Carlo tests on 2006 cover and plot frequency data. Treatment, temporal and spatial effects on individual species and other variables were assessed by repeated-measures analysis of variance (GLM procedure, Statistica 6.0). Additional measurements of vegetation structure (overall height, dwarf shrub height, graminoid height and an index of density) were also made at each quadrat, replicating the timing of vegetation measures for the bird modelling, in spring 2003 (pre-treatment) and 2006, and summer or autumn 2002 (pre-treatment), 2003, 2004 and 2006. Data were analysed separately for spring and summer, using a mixed modelling framework in SAS v.9. Full details are in Appendix 4a.

Results - vegetationFestuca ovina, Galium saxatile, Vaccinium, Nardus, Agrostis capillaris and Pleurozium schreberi were

recorded most frequently. Grazing treatment affected both cover (F=1.50; P<0.05) and frequency (F=1.46; P<0.01) of the overall plant community. The effect on cover was primarily attributable to the low sheep only treatment (F=1.62; P=0.05). Frequency was affected by both low sheep only (F=1.57; P<0.05) and low sheep plus cows (F=1.72; P<0.05), there being no significant difference between the other two treatments. Species favoured by low sheep only compared to high sheep only (cover and frequency) included F. ovina, A. capillaris, Carex binervis and Juncus squarrosus. Treatments with cows tended to have more bare peat and less litter was exposed under low sheep only.

There were no treatment effects on cover, local frequency or grazing index of any individual species or other variables analysed with the exception of Molinia. However, there were temporal and spatial effects. Overall mean species richness was 13.4 m-2 in 2002 and 13.8 m-2 in 2006, with no significant change detected. Cover of Nardus declined after 2004 from 21% to 16% (F1.9,11.5=25.4, P<0.001) as did its frequency (47% to 37%; F1.8,10.8=13.2, P<0.01). The grazing index was very low, albeit significantly higher in 2006 (4%) than previous years (0.2-1.2%) (F2.3,13.9=13.8, P<0.001). F. ovina cover was slightly higher in 2004 (14%) than in other years (11%) (F1.4,8.6=10.5, P<0.01) as was that of A. capillaris (F1.9,11.5=19.6, P<0.001). Molinia cover showed a similar trend that was just

SID 5 (Rev. 3/06) Page 16 of 29

-25000

-20000

-15000

-10000

-5000

0

-20000

-10000

0

10000

20000

30000

40000

non-significant (F1.2,7.0=5.2, P=0.05). Molinia cover and frequency were very low (1-2% and 3-4% respectively). However, the Molinia grazing index differed among treatments, being lowest in the high sheep only treatment (7%), with an increasing trend from low sheep only (21%) to cows only (46%) and low sheep plus cows (63%) (F3,3=36.3, P<0.01). Therefore, Molinia was selectively grazed, especially when cows were present, and despite being relatively scarce in the sward. The index was also significantly lower in 2004 than in other years (F2.4,7.3=10.5, P<0.01), mirroring the trend in top cover. Cover of J. squarrosus and C. binervis showed the reverse temporal pattern to F. ovina, A. capillaris and Molinia, being lower in 2004 (F1.5,8.9=5.6, P<0.05; F2.0,11.9=5.5, P<0.05 respectively). Vaccinium had relatively low cover (8-9% overall year means) but high frequency (67-72%), neither of which changed significantly over time. Vaccinium was widely grazed (15-17%) especially in 2003 (27%; F1.9,11.4=6.3, P<0.05). Calluna was scarce in the sward (cover 1-2%, frequency 2%) and not heavily grazed (3-10%). Bare ground was significantly more frequent in 2003 (3.5%) than in other years, when frequency was negligible (F1.1,6.3=19.2, P<0.01). Graminoid heights declined under the higher stocking densities (high sheep and low sheep plus cows) but increased under the lower stocking densities (cattle only and low sheep). Vegetation density in summer increased under all treatments, but more markedly under the two with higher stocking densities.

The overall grazing effect on the plant community was on both frequency and cover, indicating some change in community structure as well as differences in the amount of shoot biomass. However, the effects were mostly small in magnitude, reflecting the low dynamism of this vegetation. There was more temporal and spatial variation. In 2004, cover of productive species (F. ovina, A. capillaris, Molinia) was higher and less productive species (J. squarrosus, C. binervis) was lower. This suggests a more productive year, supported by trends in vegetation height. For Molinia at least, this was also related to reduced grazing intensity. Vaccinium showed no overall change, despite being widely grazed. Block effects were probably attributable to inherent spatial variation in the vegetation as there was no evidence that starting conditions due to the previous grazing regimes affected the vegetation dynamics.

Results - livestock Animal performance varied more between years than between treatments. There was no significant effect of

treatment on daily liveweight change (DLC) of cattle (P>0.05), although there were consistent trends. In the first year (2003), which was the first time that cattle had been introduced to these areas for many years, there was high biomass of vegetation, which had been neglected by sheep, and cattle performance was very similar (P>0.05) on both treatments (0.49 vs. 0.54kg day-1 for cattle only and mixed grazing respectively). In subsequent years, cattle performance was consistently c. 25% lower under mixed grazing. There was an effect of year (SED=0.07, P<0.001) with gains in 2005 being higher than in other years. This is consistent with the vegetation results. With more palatable species being present in the sward in September 2004 and a very warm summer with adequate rainfall, coupled with lower sward heights it would be expected that more palatable, leafy and nutritious vegetation would have been available, increasing nutrient intakes and leading to improved performance. Overall, gains in cattle liveweight were sufficient for growing heifers, calving at 3 years old.

There was no effect of treatment on lamb weaning weights (P>0.05) although there was a significant year effect (SED=0.73, P<0.001). Lamb performance appeared to be improving year on year (24.1, 25.5, 27.4 kg respectively), which again would be expected from the small but significant vegetation changes recorded.

There was a significant interaction between year and treatment on ewe weaning weights (SED=1.40, P<0.05). In 2006, ewes on the low stocking rate had a significantly greater liveweight at weaning than ewes on the high stocking rate (41.0 kg vs. 38.1 kg respectively), with ewes on the mixed grazing regime having intermediate values. In previous years, there was a similar trend (P>0.05) between ewes on the two sheep only treatments.

5.2 Redesdale System StudyThe study was carried out on 103 ha of degraded wet heath at 260-350 m asl on amorphous peat and silty

clay loam with impeded drainage. The vegetation comprised a mosaic of plant communities, the most prominent being a degraded form of the Scirpus cespitosus - Erica tetralix (M15) community (Rodwell 1991) The area had been split into two paddocks and, from 1995-2002, grazed by Scottish Blackface sheep at 1.5 and 0.66 ewes ha -1

(minus 25% October-February). The management objective was to reduce the abundance of Molinia to facilitate the restoration of dwarf shrubs and the mosaic of wet heath plant communities. From 2003-2006, the paddocks were split again and sheep grazing regimes continued but non-lactating, autumn calving Simmental X Holstein and Belgium Blue X Holstein mature cows grazed from mid-June each year. Grazing treatments were low sheep (LS; 0.66 ewes ha-1), low sheep plus cows (LSC; 0.66 ewes plus 0.75 cows ha -1), high sheep (HS; 1.5 ewes ha-1) and high sheep plus cows (HSC; 1.5 ewes plus 0.75 cows ha-1). Cows were removed at the first signs of their grazing Calluna in mid-summer and vegetation condition checked by measuring the length of five laminae of Molinia at 30 random locations per paddock. Liveweights and condition scores of cows were recorded at the start and end of the grazing period. All ewes were weighed and condition scored at tupping and pregnancy scanning. Birthweights, marking weights and weaning weights were recorded for all lambs. All livestock data was analysed using REML (Genstat).

A grid of 196 1 x 1 m quadrats at 75 m spacing was superimposed on the site. Records were made in each quadrat of presence and top cover of all plant species, litter and bare ground; local shoot frequency (presence in each 10 cm cell) and grazing index (% of occupied cells grazed) of Calluna and Molinia. Records were taken in 2001 before new grazing regimes were introduced and at intervals from 2003 to 2006 during July-August. Plant

SID 5 (Rev. 3/06) Page 17 of 29

communities were identified by detrended correspondence analysis (DCA) and fuzzy clustering (Equiha 1990). Treatment, temporal and spatial effects on individual species and other variables were assessed by multivariate repeated-measures analysis of variance (GLM procedure, Statistica 6.0). Vegetation structure data were also collected as at Pwllpeiran, in autumn 2002, spring, summer and autumn 2003, and spring and summer 2006 and analysed using a mixed modelling framework in SAS v.9. Full details are in Appendix 4a.

Results - vegetationSix plant community types were identified with dominant species Calluna (Cv), Calluna plus Molinia (CvMc),

Molinia (Mc), C. nigra (Cn), Nardus (Ns) and Juncus spp. (Js), respectively. These were variants of M15 Scirpus cespitosus – Erica tetralix wet heath, U5 Nardus stricta – Galium saxatile grassland and M23 Juncus effusus/acutiflorus – Galium palustre rush pasture (Rodwell 1991; 1992).

Molinia cover declined in both mixed grazing paddocks (29 to 18% in 2003 in LSC; 27 to 11% in 2003 in HSC), but in the Mc and CvMc types only (year x community type LSC F20,127=3.58, P<0.001, HSC F20,127=3.72, P<0.001). In LSC, frequency also declined between 2003 and 2004 (F4,37=7.62, P<0.001) and there was a non-significant downward trend in HSC (F4,37=2.57, P=0.054). The grazing indices increased substantially after introduction of cows, from 0.14 to 0.66-0.78 in following years in LSC (F4,22=89.7, P<0.001) and from 0.25 to 0.76-0.87 in HSC (F4,13=40.1, P<0.001). Grazing indices were higher in the Mc type than the Cv type in both mixed grazing paddocks (LSC F2,25=26.2, P<0.001; HSC F2,16=10.3, P<0.01). In contrast, Molinia cover increased in both sheep only paddocks, from 24% (2001) to 33% (2003) in LS (F4,39=4.08, P<0.01) and from 30% (2001-2003) to 43% (2006) in HS (F4,45=4.82, P<0.01). Frequency also increased in HS (F4,45=4.69, P<0.01) and in the CvMc type in LS (F20,130=1.82, P<0.05) but there were no consistent temporal trends in the grazing indices. No differences were detected in the grazing index among community types in either sheep-only paddock. Molinia lamina lengths were similar each year at the end of the cow grazing period in LSC but in HSC they were slightly longer in 2004 (F3,116=5.10, P<0.01), coinciding with higher summer rainfall.

Calluna cover declined in the Cv (83 to 54%) and Cv/Mc (34 to 16%) types in LSC (F20,127=2.53, P<0.001) and there was a non-significant reduction in frequency (F15,108=1.55, P=0.08). A corresponding increase in dead Calluna cover indicated the decline was due to trampling of older plants rather than grazing. Frequency also declined in the Cv type in HSC (F9,63=3.05, P<0.01). The grazing index was higher in the Mc type than the Cv and CvMc types in LSC (F2,16=15.1, P<0.001). Very little grazing of Calluna was recorded in LS and its frequency increased after 2003 (F3,40=4.97, P<0.01). Cover also declined in HS in the Cv and Cv/Mc types (F20,150=3.35, P<0.001). Calluna seedlings were counted in quadrats but were too sparse for analysis. Distribution of seedlings tended to be clumped spatially and relatively few survived from one year to the next. Most were found in mixed grazing paddocks in Cv and CvMc types and least in HS.

Changes in cover of other species were relatively minor and reflected annual variation in rainfall and temperature. Moss cover increased after 2001 in both mixed grazing paddocks in the CvMc and Mc types (LSC F20,127=1.99, P<0.05, HSC F20,127=2.15, P<0.01). Plant litter also had lowest cover in 2001 (1.2 and 0.2%) and highest in 2003 (8.0 and 6.6%) when cows were first introduced, primarily in the Mc, CvMc and Cv types (LSC F20,127=1.76, P<0.05, HSC F20,127=2.09, P<0.01). In contrast, plant litter cover in the sheep only paddocks was highest in 2006 (LS F4,39=9.64, P<0.001, HS F4,45=11.1, P<0.001).

Graminoid height and vegetation density (spring and summer) and overall vegetation height (summer only) declined more on the two mixed grazing paddocks than the sheep only paddocks, except for density in spring which showed most marked declines in HS and LSC. Non-significant trends in dwarf shrub height suggested a similar but less pronounced effect of mixed grazing within this timescale.

The most important effect was the increase in grazing of Molinia, with reduction in its cover and exposure of the moss and litter layer after cows were introduced. However, neither Calluna nor any other species had spread apart from Juncus spp. in HSC. Therefore, there was no evidence of restoration after four years of mixed grazing, although suitable conditions appeared to have been created. The reduction in Calluna by trampling in LSC could stimulate regeneration in the longer term; seedlings were concentrated in areas with established Calluna. This contrasts with the sheep-only paddocks where Molinia increased (especially in HS) at the expense of Calluna cover. However, Calluna frequency was increasing under the low grazing levels in LS, which might be beneficial in the longer term.

Results - livestockWith the exception of 2005, cattle from the HSC treatment had lower daily liveweight change than cattle on the

LSC treatment (SED=0.21, P<0.05). The higher sheep stocking decreased cattle DLC by an average of 210g. There was considerable variation between years and in 2006, cattle on LSC treatment had DLC of 1.19 kg day -1

compared to HSC of 0.64 kg day-1. Gains in cattle liveweight were sufficient for non-lactating cows. However, grazing periods were shorter after the first two years, being 9, 10, 6 and 4 weeks in 2003-2006 respectively. This was attributed to the biomass available at the start of each season and throws doubt on the sustainability of an annual mixed grazing regime at these stocking densities.

Lambs were weaned at an average of 127 days. Weaning weights were significantly higher for lambs from the LS treatments whether cattle were present or not (SED=0.67, P<0.001). Growth rates from birth to weaning were 236, 222, 208 and 211g day-1 for treatments LS, LSC, HS and HSC respectively (P<0.001). Daily liveweight gain from birth to marking at an average 25 days of age (361, 336, 316 and 299g day -1 respectively (P<0.001) suggested that much of this difference was due to increased milk yields of ewes. The inclusion of cattle had little

SID 5 (Rev. 3/06) Page 18 of 29

effect on lamb performance with the exception of 2006 when there was an apparent reduction in performance at the lower stocking rate.

Sheep stocking rate had a significant effect on ewe tupping weight in all years (SED=0.68, P<0.001) with an average reduction of about 6kg per ewe from the higher stocking rate. The effect of cattle grazing with ewes was small overall but there were indications that ewe performance was improving with time when grazing with cattle. Condition scores were not significantly different between groups (P>0.05) although a trend suggested lower scores at the higher stocking rate (2.88, 2.89, 2.70 and 2.70 for LS, LSC, HS and HSC respectively). Treatment effects on liveweight and condition score were still apparent at subsequent scanning (52.7, 52.2, 47.8, 48.3kg (P<0.001) and 2.64, 2.67, 2.46, 2.48 (P<0.05) respectively).

5.3 Impact of Grazing Regimes on Invertebrate CommunitiesThe rapid turnover of individuals within a typical invertebrate population, as well the high species richness,

sometimes results in a relatively fast response to changes in the environment. In this study particular emphasis was placed on those invertebrates likely to respond most rapidly to changes caused by different grazing management regimes. Alteration in grazing management will initially result in changes to vegetation structure, and subsequently the plant species composition; some groups of invertebrates are as sensitive to variation in structure as composition, and this, as well as the practicalities of field sampling, guided the choice of invertebrates sampled.

Identical sampling regimes were used at both Redesdale and Pwllpeiran, with the use of pitfall traps, collected at approximately 6 week intervals during late spring and summer. Pitfall traps are an ‘activity-density’ measure, and do not provide an absolute estimate of the population size, being affected by both the density of the ground layer, and the behaviour of the invertebrate. Nevertheless, they have been successfully used in this type of survey in the past. Data analysis was via standard multivariate analyses, including detrended correspondence analysis (DCA) and canonical correspondence analysis (CCA) to summarise the multi-attribute data, with Procrustes rotation used to compare different ordinations. Collection of vegetation composition and structure data are described in sections 5.1 and 5.2 above.

PwllpeiranSamples were collected from the system study in 2003 and 2006. The major groups analysed were

Hemiptera, Araneae, Diptera and Coleoptera, with total abundance counts within each group. Groups were not identified down to the species level. Redundancy analysis of the whole fauna indicated that cattle and vegetation height were the most important determinants of invertebrate composition (Monte Carlo permutation tests). Similar findings were obtained from univariate analysis of variance of the abundance of individual taxa, with cattle, sheep x cattle interaction and vegetation (as measured by DCA axis scores) the most important variables. Several taxa occurred in significantly lower numbers with cattle grazing on its own, in particular Tipulidae and Hemiptera. The meaning of the interaction term was less clear, but suggested that in some cases the impact of cattle was less when they were used in a mixed grazing system with sheep.

RedesdaleSamples were collected from two system study paddocks in both 2003 and 2005; low sheep, and low sheep

plus cattle. Several invertebrate groups were sorted and analysed in detail: Araneae (spiders), Coleoptera (beetles), Hemiptera-Auchenorrhyncha (plant bugs), Hemiptera-Heteroptera (capsid bugs etc.), Diptera (true flies). Only Araneae were sorted for 2003, other taxa in 2005. Detailed studies on Araneae were undertaken, as these were expected to show the most rapid response to changes in management over the relatively short timescale (in ecological terms) that the experiment was running. They were found to respond to changes in both vegetation community composition and structure, with Calluna and Molinia heights being particularly important. As all Araneae are predators, the effect of plant species composition is probably an indirect one, reflecting changes in underlying environmental conditions, especially moisture content in the moss/grass layer at different location across the heft.

Detailed ordination plots for various taxa are shown in Appendix 4a. The vegetation DCA ordination axis 1 described a trend from Calluna-dominated communities, through to Molinia, and finally grass/rush communities (Holcus, Agrostis, Juncus-dominated). Vegetation DCA 1 was positively correlated with Diptera species richness (P=0.008), and weakly correlated with Tipulidae and Heteroptera species richness (P=0.05). There were few significant relationships between vegetation community composition or structure and invertebrate Shannon diversity. Grazing regime was not significantly related to the species richness or diversity of any of the invertebrate taxa. When DCA ordinations of each invertebrate taxon were compared to that for the vegetation composition using Procrustes permutation tests, all were significant. Biomass of Carabidae, Staphylinidae and Tipulidae was strongly affected by vegetation composition, as measured by the vegetation DCA axis scores, but not with grazing regime.

5.4 Impact of Grazing Regimes on Bird PopulationsTo provide some assessment of the likely reliability of the bird-habitat models described above, the abundance

of meadow pipit and skylark under different grazing regimes was recorded in 2003 and 2006 on the system study plots at Redesdale and Pwllpeiran (other species being too rare to be included). Meadow pipit abundance at Redesdale appeared to have declined by over 50% on the two sheep only treatments, but by c. 20% on the two

SID 5 (Rev. 3/06) Page 19 of 29

mixed grazing treatments. Conversely, numbers of skylark appeared to remain stable or increase slightly on the sheep only plots, but to decline by c. 30-40% on the two mixed grazing treatments.

At Pwllpeiran, where there is a replicated design and statistical inference is possible, changes in meadow pipit abundance appeared to show the reverse pattern to Redesdale, increasing by 40-200% under the two sheep-only treatments, but declining slightly under the cattle and mixed grazing regimes (P=0.0044). Non-significant (P=0.15) changes in skylark abundances reflected those at Redesdale, with an approximate halving of numbers under the cattle only treatment, whilst numbers more than doubled under the low sheep only regime.

Across both sites, observed changes in meadow pipit and skylark abundance were correlated with predictions from the final models, but only on the sheep only grazing treatments (meadow pipit rs=0.69, P = 0.059; skylark (rs=0.70, P = 0.053). Observed population trends were significantly more negative than expected on the cattle grazed plots for both species (meadow pipit, F1, 12 = 6.21, P = 0.03; skylark, F1, 12 = 6.21, P = 0.03) and predicted change was not correlated with observed change on these plots. This suggests that grazing by cattle has an additional, negative effect, at least in the short term, on both meadow pipit and skylark abundance.

Findings and conclusions from these analyses and comparisons of bird abundance on the systems studies have to be treated with caution, however, given the short time period involved, and the fact that changes were determined from two snapshot surveys (rather than trend data), where counts along the transects were often small.

5.5 Experiments To Accelerate Calluna RecolonisationNested within the system study paddocks at Pwllpeiran and Redesdale were a series of plot experiments to

establish methods to increase the cover of Calluna on over-grazed grass-dominated moorlands. In particular these experiments set out to assess:

if disturbance to create bare ground is necessary a comparison of trampling by animals or mechanical methods as a means of disturbance if application of Calluna seed is required if Calluna establishment is better under grazed or no grazing treatments if the plots are grazed, then what type of stock and what stocking rate should be applied.

At Pwllpeiran the three grazing treatments used were cattle only, low sheep plus cattle and high sheep with each grazing treatment replicated 3 times. The vegetation was predominantly a mixture of fine grasses, mainly Nardus, Agrostis spp and Festuca spp with Vaccinium.

At Redesdale the three grazing treatments used were high sheep plus cows, high sheep, low sheep plus cows. The grazing treatments at Redesdale were not replicated and thus the effect of grazing on the restoration treatments could not be assessed statistically. The vegetation was dominated by Molinia with low cover of Calluna.

At each site, a total of fifty-four 10m x 10m plots were established in September 2002. Plots were randomly assigned to one of three disturbance treatments: undisturbed, rotavation or trampling. The trampling treatment was carried out by cattle at Pwllpeiran. Five 2 year old Welsh Black heifers were penned into the plot and herded round. At Redesdale the trampling was carried out by a horse that was ridden round the plots creating bare ground. At Pwllpeiran, plots were tine rotavated and then chain harrowed to bundle up the litter. At Redesdale the plots were tine rotavated with two passes, the second at 900 to the first. This created a grid of slots about 10cm apart and lifted up squares of turf. The aim with all the disturbance treatments was to create small areas of bare ground within the vegetation giving a total of 10-15 % bare ground across the entire plot. All plots were then divided into four subplots, two of which were fenced. One each of the fenced and grazed subplots were sown with Calluna seed at a rate of 0.8g/m2 in March 2003.

Calluna and Juncus effusus presence was recorded in nine 1m x 1m quadrats in each subplot each spring. Calluna cover was recorded in four 2m x 2m quadrats in each subplot at the start and end of the project . Three Calluna plants that had established after the disturbance treatments were randomly sampled from each subplot in spring 2006 and oven dried to obtain their dried weight. The presence or absence of Calluna and J. effusus was related to the different treatments and changes over time by fitting generalized linear mixed models (GLMM; SAS 9.1 macro GLIMMIX). The dry weight of the Calluna plants and Calluna cover was analysed by fitting linear mixed models (MIXED procedure, SAS 9.1). Vegetation height and changes in plant community composition were also recorded. Further details are in Appendix 4b.

Calluna establishment, cover and morphologyAt Pwllpeiran disturbance had a significant effect on the presence of Calluna (F2,46 = 25.1; P<0.001); trampled

and rotavated plots both had a significantly higher probability of Calluna occurrence than undisturbed plots. Seeding significantly increased the probability of Calluna presence (F1,203 = 432; P<0.001). By 2006, over 80% of seeded rotavated and trampled 1m2 quadrats had at least one Calluna plant present compared with less than 15% in the unseeded rotavated and trampled quadrats (Fig. 7a). On the undisturbed plots, 15% of the seeded quadrats had Calluna plants but fewer than 1% of the unseeded quadrats had Calluna plants present. Calluna presence was greater on the cattle only grazed plots than any of the other grazing treatments. The percentage of bare ground created by the disturbance treatments also had a significant effect on Calluna presence (F1,69 = 11.7; P<0.01). The influence of bare ground on Calluna presence differed between grazing treatments. If 5% bare ground was created by either trampling or rotavating, then 3 years after sowing there was 90% chance of a

SID 5 (Rev. 3/06) Page 20 of 29

Calluna plant in the cattle only grazing treatment, 80% chance in the low sheep plus cattle grazing treatment and 50% chance in the high sheep treatment.

Most plots had less than 1% Calluna cover at the start of the experiment. By the end of the experiment, Calluna cover had increased to over 10% in subplots that were seeded and disturbed and were either ungrazed or grazed by cattle only. By 2006, Calluna cover was significantly higher on plots grazed by cattle only than by high sheep or low sheep plus cattle (F2,43 = 4.3; P<0.05); on rotavated and trampled plots than on undisturbed plots (F2,43 = 18.4; p<0.001); on fenced than on unfenced plots (F1,151 = 7.6; P<0.01), and on seeded than on unseeded plots (F1,151 = 159; P<0.001). The dry weight of Calluna plants was significantly greater under cattle only compared to sheep only (F2,31 = 7.7; P<0.01) and at fenced than grazed plots (F1,34 = 61.4; P<0.001).

(a) (b)

Fig. 7. Effect of disturbance & seeding treatments on presence of new Calluna plants at (a) Pwllpeiran & (b) Redesdale in spring 2006 (3.5 years after start of experiment). S = seeded, N = unseeded.

At Redesdale, disturbance had a significant effect on the presence of Calluna (F2,40 = 44.4; P<0.001); rotavated plots had a significantly higher probability of Calluna occurrence than trampled or undisturbed plots. Calluna presence was greater in grazed than ungrazed subplots (F1,215 = 7.9; P<0.01) and seeded than unseeded subplots, (F1,217 = 63.8; P<0.001). In seeded subplots in spring 2006 Calluna occurred in 86% of rotavated quadrats, 46% of trampled quadrats, and 22% of undisturbed quadrats (Fig. 7b).

Plots had on average less than 5% Calluna cover at the start of the experiment. At the end of the experiment, ungrazed rotavated plots had over 10% Calluna cover. By 2006, Calluna cover was significantly higher on rotavated plots than on trampled or undisturbed plots (F2,45 = 25.9; P<0.001), on fenced than on unfenced plots (F1,151 = 26.7; P<0.001) and on seeded than unseeded plots (F1,151 = 24.4; P<0.001). Calluna plants from the high sheep paddock were significantly heavier than those from high sheep plus cows (F2,29 = 4.0; P=0.029) and from fenced than grazed subplots (F1,29 = 56.8; P<0.001).

For these experiments it was concluded that the creation of bare ground is essential for Calluna establishment. Rotavating and trampling were equally successful at creating bare ground for Calluna establishment at Pwllpeiran, but rotavating was more successful at Redesdale. At Pwllpeiran, seed addition was essential, as the site lacked a viable seed bank in the soil. At Redesdale, seeding increased the number of Calluna plants establishing, but in areas with a large residual seedbank seeding was not essential. Calluna establishment was generally better in ungrazed than grazed plots. At Pwllpeiran, the cattle only grazing treatment was the most successful for Calluna establishment and it also had the heaviest Calluna plants. At Redesdale, Calluna establishment was best on the field with high sheep grazing, but this may have been because this field had a higher Calluna seedbank than the other fields, and cannot be attributed unequivocally to the grazing treatment. Juncus effusus was present at both sites, and there was concern that the disturbance treatments would cause it to spread. However, disturbance only increased the presence of Juncus to very low levels, and there was no evidence of rampant expansion.

5.6 Feeding StudiesThe objective of this part of the experiment was to investigate interactions between animal type, sward

composition and diet selection in order to improve understanding of factors influencing the impact of grazing on heather moorland, and to investigate associated impacts on herbage intake, nutrient supply and rumen fermentation characteristics.

Short-term experiments were conducted at ADAS Pwllpeiran over three years (2003 – 2005) on sites with Low (8 %), Medium (36 %) and High (61 %) percentages of cover of Calluna. There were two separate 14-day experimental sessions at each experimental site, one at the end of July, and the other at the end of September. The intention was to graze two breeds of sheep (Welsh Mountain (WM) and Scottish Blackface (SBF)) and two breeds of cattle (Welsh Black (WB) and Continental crosses (CONT)) on the plots during each session. All the experimental animals used were mature, barren females, and all had previous experience of grazing hill areas.

Each 4 ha site was characterised in detail in terms of sward composition. Separate samples of the main plant species within each experimental site were taken for subsequent chemical analysis to determine chemical composition and associated nutritive value. Diet composition was estimated using n-alkane and long-chain fatty alcohol (LCFA) profiling, and feed intake and diet digestibility determined by dosing with n-alkanes. The impact of diet choice on metabolic status was monitored through blood sampling. In vitro studies using faecal samples as

SID 5 (Rev. 3/06) Page 21 of 29

0

25

50

75

100

N S N S N S

Control Rotavated Trampled

Cal

luna

pre

senc

e (%

of q

uadr

ats)

0

25

50

75

100

N S N S N S

Control Rotavated Trampled

Cal

luna

pre

senc

e (%

of q

uadr

ats)

inocula were carried out in parallel with the in vivo work, in order to investigate species/breed differences in rumen fermentation characteristics. Representative sub-samples of Calluna and grass mix were used as substrate in a gas production experiment, with fresh faecal material from each animal used as inocula. Full details of the methods used are given in Appendix 4c.

Faecal n-alkane and LCFA concentrations were subjected to principal components analysis. For dietary components, blood metabolites and gas production parameters analysis of variance was carried out using orthogonal contrasts. Diet composition data were subject to an angular transformation prior to analysis. Diet selection was quantified using Jacob’s (1974) modification of Ivlev’s Electivity Index: D = ( r i- p i) / ( r i + p i - 2r i p i), where r is the fraction of a given plant category i in the diet, and p is the fraction of the same plant category in the sward. D varied from -1 to 0 for negative selection and from 0 to +1 for positive selection.

ResultsDespite their familiarity with the vegetation type the CONT cattle struggled to maintain body condition during

some experimental sessions, in part due to their inability to tolerate the wet weather conditions, and as a result, some data was excluded from the analysis. Problems were also encountered with the analysis of the dietary samples from the Medium site.

Principal components analysis of faecal concentrations of n-alkanes and LCFA showed separation between cattle and sheep, and between breeds of cattle and sheep (illustrated in Fig 8). At the Low site the diets of both animal species were dominated by grasses. In July the sheep diets were found to contain significantly lower proportions of fine-leaved grasses compared to the cattle diets (P<0.001). They also contained higher proportions of Eriophorum spp. (P<0.01) and C. vulgaris (P<0.01), although both of these were minor components in the diet, with C. vulgaris accounting for < 3.5%. Compared with the cattle diet, the sheep diet in September again contained more C. vulgaris (P<0.001) and Eriophorum spp. (P<0.01), plus more Juncus spp. (P<0.05), Scirpus cespitosus (P<0.05) and moss (P<0.001), but lower proportions of forbs (P<0.05). When the diets of the two sheep breeds were compared there were significant between breed differences in the majority of the dietary categories during both experimental sessions. These included C. vulgaris (P<0.01), other dwarf shrub species (P<0.05), broad-leaved grasses (P<0.05), fine-leaved grasses (July) (P<0.001), Carex spp. (P<0.05), Eriophorum spp. (September) (P<0.001) and moss (P<0.05). When selectivity indices were calculated there had been consistent selection for broad- and fine-leaved grasses by both species during both sessions.  Nardus stricta was always selected by cattle, but the response of the sheep was more variable.  Although C. vulgaris had been avoided by all groups, the strength of avoidance was less for the SBF.  Other dwarf shrubs were also avoided, but to varying degrees. Carex spp., Eriophorum spp. and Juncus spp. were avoided in July while forbs were selected, but there was much more variation, both between different animal types and at the individual animal level, in the equivalent responses in September.

-2500 -2000 -1500 -1000 -500 0 500 1000 1500-500

0

500

1000

Component 1

Com

pone

nt 2

SBFWMWBCX

Fig. 8. Principal components 1 versus 2 of faecal concentrations of n-alkanes and long-chain fatty alcohols sampled in September from animals grazing the High site.

The sampling session in July at the High site revealed several between-species differences in diet composition, with the sheep diet containing less N. stricta (P<0.01) and Eriophorum spp. (P<0.05) and more broad-leaved grasses (P<0.001) than the cattle diet. The sheep diet also contained more C. vulgaris (P<0.05) , although again the proportions were small. There were no between-breed differences recorded for the sheep at this time, and the only difference between the cattle breeds was in the proportion of other dwarf shrubs (P<0.05).  In September the sheep diets still contained more C. vulgaris (P<0.001), whereas the cattle diets contained more Carex spp. (P<0.001). Several significant differences in categories that accounted for only minor components in the diet, such as moss, were also recorded.  While the diets of the two breeds of cattle were similar at this time, there were a number of significant between-breed differences in the sheep diets. The SBF diet contained substantially more C. vulgaris (P<0.001) and Eriophorum spp. (P<0.01) than the WM diet, and less broad-leaved grasses (P<0.05). When selectivity indices were calculated for the estimated diets there was consistent selection for broad- and fine-leaved grasses during both sessions.  N. stricta was tolerated by the sheep and selected by the cattle in July. This grass was selected by both species in September, but there was a greater variability in the

SID 5 (Rev. 3/06) Page 22 of 29

response of individual animals at this time. The response to non-grass monocots was also very variable during both sessions.  The SBF diets contained a higher proportion of C. vulgaris in September, and this was reflected in the corresponding selectivity index, but they had still exhibited strong avoidance of this sward component.  

During both sampling sessions at the Low site the dry matter (DM) intake per unit metabolic liveweight of the cattle and sheep was similar. However, the DM digestibility (DMD) of the diet selected by the sheep was significantly higher on both occasions (P<0.05). In contrast, the differences in diet composition between the WM and SBF sheep were not reflected in between-breed differences in intake at this site.  When grazing the High heather site in July there was no significant difference in the intake per unit live weight of the cattle and sheep, or in the digestibility of the diet.  However, in September the sheep had a higher intake per unit metabolic liveweight (P<0.05) and the diet consumed had a higher DMD (P<0.001). Between-breed differences in intake and diet DMD were recorded for both species during July (P<0.05). At this time the SBF sheep and the CONT cattle had the higher intake and diet DMD.   By September the results for the two cattle breeds were similar, and the WM sheep had the higher intake (P<0.05).

Compared to sheep and cows grazing sown pastures the blood urea concentrations of the animals in this study were very low. These were lower for sheep than for cattle (particularly the SBF sheep) at the High site (P<0.01), but were much lower in the WB cattle than either breed of sheep at the Low site (P<0.001). Considering the relatively low DMD values of the diets, it is most likely that low dietary protein intakes were the main cause of the low serum urea concentrations in all animals. While the serum non-esterified fatty acid (NEFA) concentrations recorded indicated a feed energy deficit for both sheep and cattle, the cattle values tended to be higher suggesting that the deficit was greater for cattle than for sheep. Overall, the blood metabolite results indicate that the intake of both energy and protein for all experimental animals was less than optimal to supply even maintenance requirements.

The differences in diet composition had little apparent impact on rumen fermentation characteristics. The gas production data indicated very few significant differences in the activity of microbial populations between different breeds of the same species.

6. Results Overview and Discussion

Spatial variationThere is strong evidence from both the modelling and empirical studies that the effects of particular grazing

regimes on biodiversity and economics vary among sites with different vegetation and physical characteristics. The effects of a particular grazing regime will also vary across a heterogeneous site because, as shown in the feeding experiments, both sheep and cattle select grasses in preference to dwarf shrubs and show subtle preferences for different plant species depending on their relative abundance and the season of grazing. Varying responses to grazing reported in the literature are also attributable to similar processes (e.g. Grant et al., 1985; Hodgson et al., 1991; Clarke et al., 1995; Welch & Scott, 1995; Hulme et al., 2002; Vandvik et al, 2005). The vegetation modelling showed that this feeding behaviour can modify competitive interactions between plant species resulting in diverse outcomes among different plant communities. In the Redesdale system study, this feeding behaviour was also manifest by a higher grazing intensity by cattle on Molinia in areas where it was most abundant. The vegetation model predicted that this would shift the competitive advantage in favour of Calluna. However, where Molinia was less abundant, it was predicted to increase under cattle or mixed grazing regimes. The maintenance of spatial variation in both plant species composition and vegetation structure is crucial for biodiversity. Spatial variation in vegetation was a major determinant of species composition for all the invertebrate groups studied. This is also important for invertebrate prey species of moorland birds identified in the literature review. The bird-habitat models also showed clearly the variation among bird species in their habitat preferences, for example in terms of dwarf shrub cover and vegetation height and density. In the absence of grazing, the vegetation model predicted that some competitive species would increase, which is likely to result in greater biomass and reduced heterogeneity. Grazing is therefore needed to maintain structural and compositional diversity, with certain grazing regimes being more likely to achieve conservation objectives at particular sites.

Effect of grazing regimes

Moorland with Molinia communitiesCattle utilise Molinia in favour of other grasses in summer (Grant et al., 1996a), and summer grazing by cattle

created the required conditions at the Redesdale system study for restoring dwarf shrubs and other heathland species. The vegetation model predicted that in areas where Molinia was abundant, summer cattle grazing would facilitate the expansion of Calluna at Redesdale and Molland, by reducing competition from Molinia. Other studies have also shown that grazing can affect plant community structure by modifying competition and its interaction with other factors (Welch & Scott, 1995). The desired outcome could be achieved by cattle alone or by mixed regimes that included low (0.66 ewes ha-1) year-round sheep grazing. Livestock performance was acceptable from an economic perspective in the mixed grazing regimes tested although both cattle and individual lamb and ewe performance were better at lower (0.66 ha -1) sheep stocking densities. Cattle would also attract additional agri-environment scheme payments under the current English Higher Level Scheme, although cattle only regimes would fare better economically. However, cattle could not be sustained throughout the summer after the first two years at the stocking densities tested. The overall economic sustainability of mixed grazing systems is also

SID 5 (Rev. 3/06) Page 23 of 29

questionable because significant numbers of stock need to be retained elsewhere on the farm outside the relatively short summer grazing period. Cattle trampling and grazing can have some negative effects, such as breaking of older Calluna plants or encouraging regeneration of Juncus spp., as demonstrated in the system study, the restoration experiments and the vegetation model outputs for Redesdale and Birkbeck. Vegetation at the Redesdale site had similarities to degraded blanket bog but results from the system study should not be extrapolated to intact blanket bog because of its sensitivity to damage from trampling. However, the restoration experiments also demonstrated positive effects of disturbance in wet heath, including trampling, which increases the establishment rate of Calluna seedlings. Cattle also had greatest impact on the invertebrate community composition but modelling predicted that the relative abundance of most invertebrate groups would decline under long-term cattle and sheep grazing at high cattle stocking densities. The implications for birds are difficult to assess, because much will depend upon the composition and structure of the vegetation when a new grazing regime is initiated. Inference from the bird-habitat models suggests that where Molinia is dominant, reducing its biomass through cattle grazing could be widely beneficial. Although red grouse were the only species correlated (negatively) with Molinia cover, other species may benefit through increased compositional heterogeneity, and reduction in the extent of tall, dense, vegetation. Cattle grazing may benefit those species requiring short, open, swards, although this was not borne out by skylark numbers at Redesdale. Possible negative effects of cattle grazing need to be investigated further.

Sheep only grazing regimes (at 1.5 and 0.66 ewes ha -1) resulted in Molinia increasing in the Redesdale system study and the vegetation model predicted that this would continue in the longer term regardless of stocking rate, often at the expense of Calluna. However, sheep only regimes were less damaging to Calluna at the lower stocking density (0.66 ha-1) in the Redesdale system study and a small increase in Calluna was predicted by the vegetation model. In other studies Calluna increased in proportion to reductions in sheep stocking densities (e.g. Hulme et al., 2002; Pakeman et al., 2003). Sheep grazing at low stocking densities is also more likely to maintain intact blanket bog (Grant et al., 1985). Calluna seedling establishment in the restoration plots was also better in the absence of any grazing. Individual lamb and ewe performance was also better at the lower (0.66 ha-1) stocking density. The feeding studies showed that all livestock tended to avoid dwarf shrubs if sufficient alternative grass forage was available, which would explain the effects of low (0.66 ha -1) sheep stocking densities on Calluna recovery. Sheep at the higher (1.5 ha-1) stocking density were least favourable to both vegetation and sheep performance in the Redesdale study. Economic modelling also showed that sheep only regimes were likely to result in negative net margins, at least for the moorland component of the farm. Positive margins might be achievable if Single Farm Payment and agri-environment scheme payments are taken into account, but only at lower (0.66 ha-1) stocking rates.

In communities where the starting abundance of Molinia and Calluna was low (<10%), the vegetation model predicted that increase in both species might be driven by their replacement of species such as Nardus, fine-leaved grasses, Eriophorum and Carex. This process was apparently not influenced by livestock grazing. This presents a dilemma for the management of similar heterogeneous sites, as the species replaced varies between communities. Thus each community represents a potentially important source of Molinia increase even if Molinia is controlled successfully in communities where it is dominant. Establishment of dwarf shrubs within these communities by seeding and light summer grazing might limit the expansion of Molinia by inducing competition, but this needs to be tested. Increases in Molinia and Calluna in situations where they are less abundant are likely to have mixed effects on birds, with some benefits probably arising from an increase in Calluna, but with reductions in Eriophorum and Carex cover likely to be detrimental.

Moorland with Nardus communitiesThe feeding studies at Pwllpeiran showed that cattle selected Nardus and also consumed more Nardus than

sheep, which has also been confirmed in other studies (e.g. Common et al., 1998). This suggests that cattle grazing might be beneficial where the conservation objective is to reduce Nardus and increase the diversity of other plant species including dwarf shrubs and grasses (Grant et al., 1996b). In the Pwllpeiran system study, grazing by cattle tended to expose the substrate more than grazing by sheep and Calluna establishment was best in the cattle only grazing regime in the restoration plots. In addition, performance of cattle tended to be better in the absence of sheep, although still satisfactory with sheep. Continental breeds of cattle performed satisfactorily at Redesdale but they did not cope as well with conditions in the feeding studies at Pwllpeiran compared to the traditional breed. However, the vegetation model predicted that at Pwllpeiran, cattle only or mixed grazing regimes would result in a slower decline in Nardus than would low sheep grazing (0.66 ewes ha-1). This might be explained by cattle, especially at higher stocking densities, removing substantial biomass of broad-leaved grasses as well as Nardus, so reducing interspecific competition and allowing Nardus to persist for longer in the absence of other competitive species. The feeding studies showed that cattle ingested more than sheep of several plant taxa besides Nardus and therefore lend support to the hypothesis that high levels of offtake by cattle would tend to negate any effect of feeding on Nardus at these stocking rates. Reduction in Nardus might only be achieved by even more intensive grazing of Nardus itself (Grant et al., 1996b). At Birkbeck, cattle provide an opportunity for the expansion of dwarf shrubs and rushes by reducing the occupancy of broad-leaved grasses in bent/fescue grassland. An expansion of Calluna was not predicted for Pwllpeiran as its starting abundance was very low compared to Birkbeck. Marked increases in rush abundance were predicted for Pwllpeiran, suggesting that if present, other species could exploit the removal of broad-leaved grass biomass predicted under summer cattle grazing. Therefore, opening of the sward by cattle could enhance Calluna regeneration, but only if it were present

SID 5 (Rev. 3/06) Page 24 of 29

in the seed bank or added artificially, as demonstrated clearly in the restoration plots. Without a suitable seed source, Nardus might simply be replaced by other grasses or rushes. Timing of cattle grazing on Nardus could be important, however. In both modelling and empirical studies, grazing by cattle was not done before July. It is possible that cattle grazing earlier in the summer might reduce the competitive ability of Nardus.

An increase in Nardus has been observed widely across the UK uplands where sheep have been grazed at stocking densities in excess of 2 ewes ha-1 (Thompson et al., 1995). Selection by sheep of both fine- and broad-leaved grasses (Grant et al., 1985b), confirmed in the feeding studies, would explain this response of the less palatable Nardus. At low sheep densities (0.66 ha-1), model predictions suggested that an increase in broad-leaved grasses could reduce Nardus as a result of competition. An increase in palatable grasses alone would not necessarily be beneficial in terms of conservation objectives but the model predicted that other species such as Vaccinium would also increase. The effect on palatable grasses was partly confirmed by the Pwllpeiran system study in which they tended to be more associated with the lower sheep stocking density (1.0 sheep ha -1). At low sheep stocking densities it appears that offtake of other grasses is insufficient to give Nardus a competitive advantage, particularly if the favoured grasses form a major component of the sward, as in the Pwllpeiran system study and the bent/fescue community at Birkbeck. Cattle grazing resulted in a reduction in Tipulids, which are widely reported in the literature as being important prey items for moorland birds. Nardus cover was not associated with the abundance of any of the bird species studied (although it was negatively associated with occurrence of merlin). Therefore, few implications can be made from changes in Nardus vegetation, although low sheep stocking densities that cause a reduction in Nardus and a concomitant increase in plant species such as Vaccinium might be beneficial overall to birds through increased compositional heterogeneity. Stocking density had more impact on vegetation structure than cattle grazing at Pwllpeiran and might be the best way of managing for bird species associated with different structures. Ewe performance was better at the lower sheep only stocking rate at Pwllpeiran although lamb performance was similar. Low sheep only grazing regimes are therefore more likely to meet the conservation and economic objectives than cattle grazing in Nardus moorland.

In summary, both positive and negative outcomes are predicted for summer cattle or low sheep (0.66 ewes ha 1) grazing on moorlands with Nardus communities and the final choice will be a balance between effectiveness in delivering management objectives and economic sustainability. At Pwllpeiran, low (0.66 ha -1) sheep only regimes appear more likely to meet the conservation and economic objectives than cattle grazing. At Birkbeck, a rotational regime of summer only cattle (for say 1-2 years) with low (0.66 ha-1) all year sheep grazing might enable an initial expansion of dwarf shrub species and the retention of sufficient broad-leaved grass cover to out-compete Nardus. However, the restoration experiments demonstrated seed addition is necessary to regenerate Calluna if it is deficient in the soil seedbank.

Rates of changeSuitable conditions had been created under the mixed grazing regimes at the Redesdale system study after

four years, with reduction in Molinia cover and exposure of the moss and litter layer but there was little evidence of restoration of dwarf shrubs and other plant species. Similarly, there was little vegetation change in the Pwllpeiran system study apart from annual fluctuations in cover that corresponded to variation in rainfall and temperature. Vegetation change at the case study sites was also slow. Invertebrates with strong host plant specificity (Auchenorrhyncha) were likewise little affected by grazing regimes after four years, presumably because there had been few changes in plant species composition. This concords with predictions from the vegetation model, which showed only slow changes over a 20-year period, although it did suggest that change might be evident in the Molinia dominated and mixed heath communities at Redesdale after about five years of summer cattle grazing at high stocking densities. The dynamics of species change was also predicted to vary considerably between different plant communities and may be driven by different processes. The restoration experiments demonstrated how regeneration of Calluna can be accelerated by disturbance, seed addition and at least temporary cessation of grazing.

7. Conclusions1. Vegetation responses to particular grazing regimes will vary among sites and among plant communities within

individual sites. Habitat requirements, and therefore the response to a grazing regime, also vary among bird and invertebrate taxa. Diverse vegetation structure and plant species composition within and between sites are required to maximise biodiversity. It is not possible, therefore to produce generalised grazing prescriptions but specific grazing regimes are most likely to meet individual site objectives for biodiversity and economics. Practical guidelines are in Appendix 5.

2. Summer grazing by cattle is potentially beneficial for moorland regeneration by reducing competitive grasses. Cattle are particularly effective for reducing Molinia but their effectiveness for managing Nardus appears to be more variable and dependent on plant community composition. Performance of both traditional and continental breeds can be satisfactory although welfare issues can arise with continental breeds under inclement weather conditions. Positive economic returns can be achieved, depending on scheme payments received. However, uniform heavy grazing by cattle is likely to reduce invertebrate abundance. Cattle grazing might have some benefits for birds although varying stocking density might be the best way of managing for bird species associated with different structures. High stocking densities are unlikely to be sustainable and provision for long periods off the moorland will be needed. Trampling by cattle can be beneficial for dwarf shrub re-establishment but potentially damaging to older Calluna and to intact blanket bog.

SID 5 (Rev. 3/06) Page 25 of 29

3. Historically, sheep grazing at high stocking densities have resulted in a reduction of dwarf shrubs and replacement by other species such as Molinia and Nardus. At low stocking densities, sheep are much less damaging to both established Calluna and new seedlings and can also reduce Nardus by facilitating competition from broad-leaved grasses. This might be beneficial overall to birds through increased compositional heterogeneity. Invertebrates are associated with variation in plant species composition and vegetation structure but responses to sheep grazing regimes are not easily predicted. Individual lamb performance also tends to be better at low stocking densities. Sheep only regimes are only likely to achieve a positive economic return at the moor level at low stocking densities with agri-environment payments.

4. It is possible to achieve at least some economic and environmental objectives simultaneously under mixed or single species grazing regimes, depending on the plant communities present and as long as stocking densities are not high. However, varying habitat requirements of different taxa will limit what can be achieved under a particular grazing regime at individual sites.

5. Restoration of degraded moorland by grazing alone will be slow (decades) although this will vary among plant communities. Localised disturbance by rotavation or trampling can accelerate Calluna re-establishment in the first three years. Seed addition will be necessary if there is no seedbank and can increase the probability of establishment by up to fivefold if combined with disturbance.

6. The field experiments carried out in this study were short-term and therefore focussed on the re-establishment and early restoration phase in degraded moorland habitats. These were complemented by longer-term predictions from models. The vegetation model performed well against short-term field data, but its accuracy over longer timescales is untested. Linkage of the vegetation model outputs to the bird models was not successful because the current version does not include outputs on vegetation structure. Vegetation structure outputs could also improve predictions from the invertebrate models. The vegetation model could also be improved by inclusion of trampling effects. The timing of more subtle adjustments to grazing regimes, for example the reintroduction of grazing following Calluna seedling establishment is also untested. Economic models were run using payment levels from English policy schemes, so predicted outcomes could vary under different policy scenarios, including those in Wales.

AcknowledgementsThis project was funded by Defra, English Nature (now Natural England) and the Countryside Council for Wales. We are grateful to the project Steering Group: Richard Brand-Hardy, Katherine Bass, Mark Baylis, Emma Clare, Ian Condliffe, Alistair Crowle, Ann Davies, Richard Evans, David Garwes, David Glaves, Belinda Gordon, Barbara Jones, Mick Rebane, Mike Roper and Lesley Stubbings.

SID 5 (Rev. 3/06) Page 26 of 29

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

SID 5 (Rev. 3/06) Page 27 of 29

Journal papersBuchanan, G.M., Pearce-Higgins, J.W. & Grant, M.C. (2006) Observer variation in estimates of Meadow Pipit (Anthus pratensis) and Skylark

(Alauda arvensis) abundance on moorland. Bird Study 53: 92-95.Buchanan, G.M., Grant, M.C., Sanderson, R.A. & Pearce-Higgins, J. W. (2006) The contribution of invertebrate taxa to moorland bird diets and

the potential implications of land-use management. Ibis 148: 615–628.Fraser M.D., Theobald V.J. & Moorby J.M. (2006) Determining the diet composition of animals grazing complex swards using n-alkanes and

long-chain fatty alcohols. Ecological Applications 16: 1901-1910.Parveen, I., Moorby, J.M., Fraser, M.D., Allison, G.G., Kopka, J. (2007) Application of gas chromatography/mass spectrometry metabolite

profiling techniques to the analysis of heathland plant diets of sheep. Journal of Agricultural and Food Chemistry, 55, 1129-1138Parveen, I., Moorby, J.M., Hirst, W.M., Morris, S.M., Fraser, M.D. (In press) Profiling of plasma and faeces by FT-IR to differentiate between

heathland plant diets offered to zero-grazed sheep. Animal Feed Science and TechnologyConference proceedings, abstracts & other articlesAllison, G.G., Fraser, M.D., Moorby, J.M., Kopka, J., Erban, A., Theobald, V.J. (2005) Metabolic profiling of heathland plants in the diet of

sheep. Proceedings of the XXth International Grassland Congress, eds. F.P. O’Mara et al., Dublin, p. 494.Buchanan, G.M. (2006) Invertebrates in the diets of moorland breeding birds. Conservation Science in the RSPB 2006, pp 35-36.

http://www.rspb.org.uk/Images/ConScifinalreport_tcm9-137443.pdf.Critchley, C.N.R. (2004) Environmentally sustainable and economically viable moorland grazing systems. The Heather Trust Annual Review

2004, 30-31.Critchley, C.N.R. (2005) Sustainable grazing of moorland. ADAS Science Review 2003-2004, 12.

http://www.adas.co.uk/media_files/Publications/science_review_2003-2004.pdf.Critchley, C.N.R., Adamson, H.F. & Hyslop, J.J. (2005) Short-term impact of sheep and cattle grazing on upland wet heath vegetation.

Proceedings of the British Society of Animal Science 2005, 230. BSAS, Penicuik.Critchley, C.N.R., Adamson, H.F. & Hyslop, J.J. (2005) Short-term impact of sheep and cattle grazing on upland wet heath vegetation. Abstracts:

British Ecological Society, Annual Meeting, University of Hertfordshire, 5-7 September 2005, 45.Davies, D.R., Fraser, M.D., Theobald, V.J. & Brooks, A.E. (2004) Characterising the fermentation capabilities of gut microbial populations from

cattle and sheep grazing heathland forage using gas production. Reproduction Nutrition Development, 44 (Supplement 1), S72.Davies, D.R., Fraser, M.D., Bakewell, E.L. (2005) Characterising the fermentation capabilities of gut microbial populations from different breeds

of cattle and sheep grazing heathland. In: Pastoral Systems in Marginal Environments, ed. J.A. Milne, Proceedings of a satellite workshop of the XXth International Grassland Congress, Glasgow, p. 157. Wageningen Academic Publishers.

Fraser, M.D., Moorby, J.M., Theobald, V.J., Jones, R. (2005) An evaluation of the n-alkane technique for determining diet composition in animals grazing complex swards. Proceedings of the XXth International Grassland Congress, eds. F.P. O’Mara et al., Dublin, p. 545.

Gardner, S.M., Buchanan, G.M., Pearce-Higgins, J.W. & Grant, M.C. (2005) Predicting the effects of grazing management on moorland bird abundance. Proceedings of the British Society of Animal Science 2005, 69. BSAS, Penicuik.

Gardner, S.M., Buchanan, G.M., Pearce-Higgins, J.W., Grant, M.C. & Waterhouse, A. (2005) Predicting the effects of management on upland birds, economy and employment. In: Pastoral Systems in Marginal Environments, ed. J.A. Milne, Proceedings of a satellite workshop of the XXth International Grassland Congress, Glasgow, p. 179, July 2005.

Grant, M., Pearce-Higgins, J., Buchanan, G. & Sim, I. (2006) Relationships between moorland birds and vegetation condition: implications for the role of management. In: Sustainable Uplands Conference: Future Scenarios for People, Environment, and Landscape: A Future Vision Conference and workshop for the Uplands, The International Centre for the Uplands – Cumbria, p. 23.http://www.theuplandcentre.org.uk/Reference/2020vision/Abstracts/Abstract%20Book.doc.

McLean, B.M.L., Davies, O.D, Griffiths, J.B., Evans, D.E. & Clarke, A. (2005) Short-term impact of grazing prescriptions on cattle performance. Proceedings of the British Society of Animal Science 2005, 234. BSAS, Penicuik.

Parveen, I., Moorby, J.M., Hirst, W.M., Morris, S.M., Fraser, M.D. (2006) Profiling of plasma and faeces by FT-IR to differentiate between heathland plant diets offered to zero-grazed sheep 8th British Grassland Society Research Conference, 'Grassland as a Multifunctional Resource', Royal Agricultural College, Cirencester, 4-6 September 2006. p 69  101-102

Parveen, I., Moorby, J.M., Fraser, M.D., Allison, G.G., Kopka, J. (2006) Application of GCMS-TOF metabolite profiling techniques to the analysis of heathland plant diets of sheep 8th British Grassland Society Research Conference, 'Grassland as a Multifunctional Resource', Royal Agricultural College, Cirencester, 4-6 September 2006. p 103-104

Paterson, L., Sanderson, R.A. & Rushton, S.P. (2005) The effects of grazing on spider assemblages in upland heather moorland. Proceedings of the British Society of Animal Science 2005, 65. BSAS, Penicuik.

Theobald, V.J., Fraser, M.D., Moorby, J.M. (2005) Effects of breed and stage of growing season on the metabolic profile of sheep grazing moorland. In: Pastoral Systems in Marginal Environments, ed. J.A. Milne, Proceedings of a satellite workshop of the XXth International Grassland Congress, Glasgow, p. 156, July 2005.

Waterhouse, A. (2006) Predicting the effects of management change on moorland vegetation, birds, economy and employment. 8-11. In: Looking to the Hills, Issue 14. JNCC. http://www.jncc.gov.uk/page-1844

Waterhouse, T., Gardner, S., Pearce-Higgins, J. & Grant, M. (2006) How might CAP reform influence land management, moorland vegetation, birds, economy and employment? In: Sustainable Uplands Conference: Future Scenarios for People, Environment, and Landscape: A Future Vision Conference and workshop for the Uplands, The International Centre for the Uplands – Cumbria, pp 24-25.http://www.theuplandcentre.org.uk/Reference/2020vision/Abstracts/Abstract%20Book.doc

SID 5 (Rev. 3/06) Page 28 of 29

SID 5 (Rev. 3/06) Page 29 of 29