NatureScot Research Report 1394 Consolidation of previous investigations into the range, variation and interpretation of herbivore impact assessment methodologies currently in use in Scotland and distillation into a user-friendly practitioners’ guide

Published: 2026 
Authors: Smith, S. W¹., Hewison, R.L.¹, Potts, J.M.², Kyle, C.¹, and Pakeman, R.J¹ 

¹The James Hutton Institute, Scotland, UK, ²Biomathematics and Statistics Scotland, UK

Cite as: Smith, S. W¹., Hewison, R.L.¹, Potts, J.M.², Kyle, C.¹, and Pakeman, R.J¹  Consolidation of previous investigations into the range, variation and interpretation of herbivore impact assessment methodologies currently in use in Scotland and distillation into a user-friendly practitioners’ guide and technical annex. NatureScot Research Report 1394.

Smith, S. W¹., Hewison, R.L.¹, Potts, J.M.², Kyle, C.¹, and Pakeman, R.J¹ 

¹The James Hutton Institute, Scotland, UK, ²Biomathematics and Statistics Scotland, UK

Keywords

Deer management; Herbivore Impact Assessment; indicators; impact classes; open hill range; sampling design; stakeholder consensus; surveyor consistency 

Background

NatureScot commissioned this project to consolidate information on the current range of Herbivore Impact Assessment (HIA) methods used across Scotland for a range of purposes and between practitioners, and to gain consensus to aid more informed application in the future. 

The overall aim of the project was to produce a readily accessible document that guides practitioners on the various methods of HIA available, the variations on standard methods currently used and their purposes. The report should include guidance on how results should be interpreted and reference to desired outcomes. Compatibility and read across between the results of the methods should also be investigated.

The guide needed to be a pragmatic method of establishing the current level of grazing/trampling pressure, and one capable of detecting temporal change in herbivore pressure on open hill habitats. This work was to focus solely on upland open range habitats in Scotland. Woodland habitats were specifically excluded from the remit of this work as other work has regularly refined the Woodland HIA, although discussions with NatureScot and stakeholders highlighted the need to consider cross-over of indicators in the Woodland HIA guide specifically where there are woodland expansion objectives. 

Specified deliverables as part of the contract were the following: 

• To collate information on existing methods for HIA and the variances between them
• Rationalise HIA attributes to remove uninformative or difficult to apply ones
• Seek consensus on the most appropriate methods and attributes to use – with the involvement of key stakeholders (see below)
• Create a user’s guide to HIA methods, with a suggested revised attribute table to enable the creation of field forms and results analysis and interpretation
• Create a technical appendix detailing the basis for the above guidance and suggested revisions (testing and adoption will follow in later phases of the project, however stakeholder feedback will be sought on the revised provisional guidance)

Main findings

• Rationalising attributes and cross tabulation of Herbivore Impact Assessment (HIA) indicators identified limited overlap in indicators across existing HIA methods. For 76 indicators used across four habitat types, out of five HIA methods, only 7 indicators were shared across four or more HIA methods and 24 were shared across more than 2 HIA methods.
• Results from an online stakeholder survey found no difference in user preference of indicators between HIA methods. However, there were significant differences in how informative users viewed HIA for different habitat types. Indicators used to assess tussock grasslands were viewed as the least informative or useful and had the strongest need for improvement.
• Synthesising the results from the stakeholder survey and consultation as part of an in-person stakeholder workshop, we have set out a list of indicators that stakeholders and the James Hutton Institute project team have confidence in for the four habitat types: blanket bog, dwarf-shrub heath, smooth grassland and tussock grasslands (specifically purple moor grass dominated tussock grasslands). We have focused on these four dominant open hill range habitats, but future work may expand upon these habitat types, especially other types of tussock grassland and considering habitats of high conservation interest.
• These indicators require field-testing and comparison to common existing HIA methods. As many of the thresholds between impact classes have limited or no scientific underpinning, we recommend that the first field-tests record raw data rather than just impact class. For example, for heather browsing reporting the impact class and estimating the proportion of shoots browsed to nearest 5%. Recording raw data would allow further adjustments of thresholds and support statistical analysis. In addition, we suggest that a correlation analysis be carried out on raw data to assess if the indicators are picking out the same patterns.
• We also suggest that there is a study of between-surveyor consistency. Observation variation can be reduced by calibrating between surveyors at the start of a survey, for instance using images of proportion of browsed shoots. Field-testing can be used to assess surveyor-drift during a survey and the consistency of thresholds and scoring. Testing should include a diverse range of users (i.e., deer stalkers/estate managers, consultants, NGOs/government agencies) to assess if the methods work for all likely user-groups.
• Statistical approaches to analysing ordinal data are not straightforward. Formal statistical analysis may not be required at the estate level, but advice needs to be developed for formal analysis on what level of change is needed for a given sample size before it can be taken as change rather than noise (sampling variance) in impact data. This would require statistical analysis of some repeat HIA survey data.
• There was no consensus amongst stakeholders on the best sampling design. Utilising georeferenced locations from past surveys, a desk-based exercise could investigate potential differences in designs between fixed point versus walk-through HIAs. We suggest selecting fixed-points on a grid-intersect to capture a representative proportion of target habitats in the sample area and using a plot-based survey approach. This may not be appropriate or economic for surveying large areas, so additional thought would be needed, potentially using the sampling approach taken in the Macaulay Land Use Research Institute’s (MLURI) Rapid Assessment of Grazing and Trampling Impacts method, which is a grid-cell and walk-through survey method. 

Acknowledgements

We would like to thank all stakeholders who answered our online survey, attended the workshop and especially those who provide constructive feedback on the draft of this report. Thank you to the members of the NatureScot Scientific Advisory Committee (SAC) Deer subgroup who provided valuable guidance and feedback on the project and report. 

Abbreviations

Common Standards Monitoring (CSM)

Deer Management Groups (DMGs)

Herbivore Impact Assessment (HIA)

James Hutton Institute (JHI)

Macaulay Land Use Research Institute (MLURI)

Wild Deer Best Practice (WDBP)

Scientific Advisory Committee (SAC) Deer Subgroup

Introduction

What is a Herbivore Impact Assessment (HIA)?

Herbivore Impact Assessments (HIAs) are field survey-based methods to assess the impact of herbivores on habitats. Assessments are based primarily on identifying vegetation features that have changed as result of herbivore presence. For Scottish open hill rangelands, assessments seek to assign a level of herbivore impacts on habitats, focusing on impacts of dominant mammalian herbivores, namely deer, sheep, cattle and to a lesser extent smaller herbivores such as hares, rabbits and voles. Indicative features of herbivore impact, or indicators, include evidence of grazing or browsing of vegetation, changes in plant growth-form and species composition, evidence of disturbance and trampling, and dung deposition. Whilst all the HIA methods attempt to assign a level of grazing or trampling, others examine the impact as well. For a given indicator, the extent of herbivore impact is then assigned an impact class. For example, widespread grazing of preferred species has high impact compared to patchy grazing of preferred species which would be assigned a moderate impact. Impact classes are often delineated between two or three classes, e.g., high, medium, and low herbivore impact. An overall herbivore impact score is then assigned to the location of the habitat type. Impact classes for each habitat type are often visualised spatially or graphically.

What are the different types of open hill range HIA in used across Scotland? 

As part of this project, we have assessed the following open hill range Herbivore Impact Assessments: 

• Pre-MacDonald method – these were earlier HIA methods used by some private estates prior to the formulation and publication of the MacDonald et al. method (Rae, 1995, Rae, 1996).
• MacDonald et al. method (also referred previously as Scottish Natural Heritage method) used by NatureScot for HIA work across Deer Management Groups (DMGs) or individual holdings to assess habitat condition (MacDonald et al., 1998a, MacDonald et al., 1998b, MacDonald, 2007).
• Wild Deer Best Practice (WDBP) designed for open hill range and woodlands and targeted to be carried out by deer stalkers and estate managers on private estates or as part of surveys coordinated across DMGs with indicators derived and simplified from the MacDonald et al. method (Wild Deer Best Practice Guidance for Scotland, 2018).
• Putman method a non-plot-based assessment methodology used by several stalkers and managers and professional contractors on Scottish estates with recent research trials in northern England (Logan et al. unpublished). The method was partly developed by Professor Rory Putman, simplifying and developing the method from the MacDonald et al. method.
• Common Standards Monitoring (CSM) informs Site Condition Monitoring of designated sites such as Sites of Special Scientific Interest. This method assesses many aspects of site condition, including herbivore impacts (Joint Nature Conservation Committee, 2009). 

Other variations and derivates of these methods were not the focus of this report. However, we did consult the Woodland HIA guide (Armstrong et al., 2023) and the Deer Initiative Woodland Impact method in England and Wales (The Deer Initiative, 2024) due to woodlands often being part of an open hill range habitat mosaic. Also, within Scotland there is an increasing interest in afforestation and reforestation with a need to understand herbivore impacts on these establishing treescapes. 

What is the purpose of HIA? 

The purpose of the different HIAs methods outlined above are diverse, albeit with a common theme of assessing herbivore impacts. A detailed, internal report commissioned by NatureScot entitled ‘Deer Management and Habitat Impact Assessment; review of the data’ (Mayne, 2021) reviews in detail the purposes of different HIA methodologies. Of the methods outlined above an exception is CSM which is designed as a pass or fail system for habitat condition of designated sites across Scotland, nevertheless, we include it in the current report as many indicators are shared. 

As part of this project, a meeting with NatureScot staff across departments (online 22^nd January 2025 with 20 attendees) identified common purposes for HIAs including: informing herbivore and deer management, understanding where management interventions are required, assessing the impact of interventions, and as evidence to support regulatory decisions. HIA assesses a specific herbivore pressure on habitats, but how it is interpreted, and hence informs management change decisions, will depend on the proposed herbivore management changes and habitat outcomes desired. 

What are the current challenges with HIA? 

Since the development of the MacDonald et al. method there have been many challenges and issues raised concerning HIAs. A main overarching concern for HIAs is a question over the robustness and consistency of results, particularly with different methodologies leading to differing results which has affected confidence in HIAs. This needs to be addressed as they can be used as the basis of regulatory approaches. Conflicting results following different HIA methods affects decision-making to inform management, interventions and regulation. Notably, conflicts arise between results from the MacDonald et al. method and the simpler WDBP, although comparisons between other methods have also resulted in different outcomes. At the same time, there is a strong desire for HIA methods to have widespread uptake by practitioners, thus simpler methods need to be as robust as more complex methods. These challenges sit within the context of changing environmental policy, with the Natural Environment (Scotland) Bill seeking to revise legislation and regulation around deer management. The Bill seeks to move away from interventions focussed on the existence or continuing of current herbivore damage to habitats towards the restoration of previously damaged habitats, that may, or may not, still be experiencing herbivore damage. 

Previous work has explored several issues concerning the robustness and consistency of HIA methods. In brief, some major issues previously studied are detailed below:

• Spatial-scale and sampling design of HIAs: this issue related to two problems (1) the repeatability of results and, (2) the time and financial cost of undertaking surveys at large spatial scales, particularly across multiple land holdings as part of Deer Management Groups (DMGs). The original MacDonald et al. method did not outline a sampling design with the assumption that this followed principles of Phase I habitat monitoring based on polygon mapping units of habitats. Variations of HIA methods differ in using randomly sampled versus fixed points and permanently marked or unmarked points. Rapid assessment methodologies were developed, such as the Macaulay Land use Research Institute (MLURI) Rapid Assessment method, sampling 0.5 km x 0.5 km squares to optimise time and expense costs of surveying large areas (MacDonald, 2007).
• As part of this report, we have sought feedback from stakeholders on preferred sampling design and reviewed different methodologies in detail. Their feedback was not conclusive, and consequently we have recommended following a fixed-point and plot-based method using GPS relocation with the option of the MLURI Rapid Assessment method for large areas such as whole DMGs.
• Surveyor consistency: A surveyor field test conducted on the MacDonald et al. method in upper Strathspey concluded complete disagreement between surveyors was unlikely across two impact classes (low versus high). In the study, consistency varied by habitat and the highest agreement was found for dwarf-shrub heath and wind clipped heath and lowest agreement for springs and flushes (McConnell Associates, 2000, MacDonald, 2010). A potential caveat of the finding was that a disagreement was only considered for a difference of greater than two herbivore impact classes, when a difference in one impact class could be considered significant disagreement. In contrast, a more recent unpublished study found high surveyor consistency within plots using the MacDonald et al. method (Headley et al. unpublished data). Similarly, a survey consistency report for Woodland HIA also found significant inter-surveyor variation for woodland methods, that led to further refinements of the descriptions and user-guide information provided for the Woodland HIA method (Armstrong, 2021).
• As part of this report, we did not investigate inter-surveyor consistency, but any revision and refinement of open hill range HIA methods would require testing inter-surveyor consistency and surveyor ‘drift’ with time.
• Streamlining indicators: The MacDonald et al. method has many indicators which are rarely recorded in HIA reports. A review of eight HIA for blanket bogs identified indicators that performed poorly with regard to their universality, sensitivity, correlation with other indicators (Pakeman, 2009). Out of 12 blanket bog indicators assessed, statistical analysis found three indicators could be omitted without loss of information and a further two indicators had only moderate support for retention (Pakeman, 2009).
• As part of this report, our focus was streamlining indicators found across HIA methods using a combination of desk-based review and stakeholder consultation. Nevertheless, we stress revisions and refinement of open hill range HIA indicators would require field testing and statistical analysis. 

Methods

Methods summary

To achieve the deliverables of the contract we combined a desk-based assessment of HIA indicators and stakeholder engagement. A summary of the approach is listed directly below with a more detailed description of methodology underneath the summary. 

• Cross-tabulation of HIA indicators: we collated all indicators across existing HIA methods targeting four habitat types (blanket bog, dwarf shrub heath, smooth grassland and tussock grassland) which were the focus of multiple methods of assessment. The purpose of this exercise was to identify shared and non-shared indicators across HIA methods by habitat type.
• Online stakeholder survey: we created an online survey for stakeholders to provide general feedback on HIA methodologies and to evaluate indicators across HIA methods for three habitat types: blanket bog, dwarf shrub heath and tussock grassland. The purpose of the online survey was to obtain wider feedback on HIA methods and indicators beyond a stakeholder workshop. Smooth grassland had a strong agreement of indicators in cross-tabulation and was not included in the online survey. We decided that asking for input on more habitats would be likely to deter stakeholders from completing the survey.
• In-person stakeholder workshop: we organised a workshop targeting invitees to represent developers of HIA methods and users from a variety of backgrounds (consultants, government agencies, NGOS, estate managers). Informed by participant presentations and results from the online survey, stakeholders at the workshop discussed issues concerning (a) HIA methods raised in the survey, (b) selection of key indicators for blanket bog, dwarf shrub heath and tussock grasslands and (c) the description of impact classes within those indicators. We focussed on these three habitats to give attendees sufficient time to consider them during the workshop.

Ethical statement

The project received ethical approval from the human ethics committee at the James Hutton Institute for undertaking an online survey and in person stakeholder workshop (reference JHI - HRE - 0304 - 573 (v2)). All participants for both survey and workshop were required to provide consent.

Cross-tabulation of HIA indicators

We reviewed documentation of five herbivore impact assessments: (1) Pre-MacDonald et al. method; (2) MacDonald et al. (1998) method (MacDonald et al., 1998a, MacDonald et al., 1998b); (3) Wild Deer Best Practice method (WDBP); (4) Putman method and (5) Site Condition Monitoring (SCM). These were reviewed for four habitat types: (1) blanket bog, (2) dwarf shrub heath, (3) smooth grassland and (4) tussock grassland. These four habitats were selected as the focus of HIA evaluation because they cover the largest area of open hill range habitats across Scotland. Further, these habitats are shared across all HIA methodologies evaluated with the exception of tussock grassland not being in the pre-MacDonald method and a tussock grassland methodology was unpublished for WDBP method and not available online to practitioners (Wild Deer Best Practice Guidance for Scotland, 2018). Other habitats covered by open hill rangeland HIA methodologies were: 

• wind clipped summit heath found in pre-MacDonald et al., MacDonald et al. and WDBP
• flush springs, tall herbs and scrub in MacDonald et al. and WDBP
• woodland in WDBP and the Putman method
• bracken in MacDonald et al., and wet heath which was differentiated from dry dwarf shrub heath and blanket bog in the Putman method. 

Indicators were tabulated for each habitat, listing all indicators as rows and HIA method as columns. Rows were further divided for each indicator listing impact classes. Indicators were considered the same across HIA methods when there was only minor variation in wording and/or interpretable description. For example, trampling due to herbivores and trampling due to deer and sheep were listed as the same herbivore trampling indicator. Indicators were viewed as different if details were not shared and when similar indicators themselves were listed as separate indicators within a given HIA method, as often occurred in the MacDonald et al. method. For example, browsing of Calluna vulgaris and Vaccinium myrtillus, summer browsing of Calluna vulgaris and browsing of older woody shoot material from Calluna vulgaris and/or Vaccinium myrtillus were treated as separate indicators in the MacDonald et al. method.  In our tabulation for the MacDonald et al. method, we collated all indicators listed under small-scale, large-scale and trend indicators sections. For the pre-MacDonald method we used the crib sheet in Rae 1995 & 1996 as our guide of indicator descriptions, since this was the oldest record, we had of a pre-MacDonald method (Rae, 1995, Rae, 1996). 

Online stakeholder survey

The aim of the online stakeholder survey was to obtain feedback on HIA methodologies and seek consensus on the value and interpretation of indicators used across HIA methods.

The survey used the platform Qualtrics. and the survey was open to respondents for six weeks between 16^th January and 28^th February 2025. A key justification in creating the survey was to receive feedback from a wider pool of stakeholders beyond those attending the in-person meeting. 

The survey focused on four commonly used HIA methods: (1) pre-MacDonald et al. method; (2) MacDonald et al. method; (3) Wild Deer Best Practice method and (4) Putman method. The survey focussed on three dominant open hill habitats: (1) blanket bog, (2) dwarf shrub heath, (3) tussock grassland. Smooth grassland was not included in the survey as this habitat had the greatest consensus of indicators across HIA methods (see results below) and to limit the complexity of the survey. The survey was advertised via social media, key stakeholder forums (e.g., Association of Deer Management Groups), and by word-of-mouth. 

As part of our evaluation of indicators, respondents were presented with a table of indicators for a given HIA method and habitat type with four selectable options to evaluate each indicator: (1) I record this indicator most frequently; (2) I rarely or never record this indicator; (3) I find this indicator useful and/or informative to herbivore impact, and (4) I find this indicator easy to confuse with non-herbivore impacts (e.g., exposure and wind clipping, burning, water-based erosion). Multiple statements could be selected for each indicator, except for selecting an indicator as both frequently and rarely recorded. If a statement was not selected for an indicator, we informed respondents that we assumed that this meant the opposite meaning of the statement, and in the case of neither frequently nor rarely recorded the indicator was assumed to have moderate usage. 

Responses to the survey on the frequency, rarity, useful/informative and ease of confusion with non-herbivore impacts for each indicator were expressed as a percentage representing the proportion of respondents selecting those categories. The analysis of proportion data used generalised linear models assuming a beta distribution for all analyses. We analysed the proportion of responses in each category in relation to HIA methods and habitat types. Across indicators we analysed the proportional response for each category in relation to one another. Finally, combining survey results and cross-tabulation of indications, we analysed the relationship between the number of times indicators were shared across HIA methods and the proportion of responses identifying the indicator as useful/informative. Proportion data was fitted in the models using the R package betareg (Cribari-Neto and Zeileis, 2010) in R (ver. 4.4.0, R Core Team 2024) (R Core Team, 2024).

Stakeholder workshop

The stakeholder workshop was an in-person event to share knowledge of different open hill range HIA methods, general challenges and potential improvements, and targeted input on indicators and impact classes. NatureScot provided a short-list of potential attendees which the project team added to. The diversity of attendees was designed to represent different HIA user groups with familiarity with different methodologies; the attendees encompassed HIA method developers, consultants, NGOs, estate managers/deer stalkers, and researchers. In advance of the workshop, attendees were requested to answer the online survey. The event was held at University of the Highlands and Islands, Inverness, on 4^th February 2025. In total 16 people attended in person with two joining online. 

The structure of the workshop was first knowledge sharing with presentations by participants on different HIA methods and some research evaluating methods and survey consistency followed by an open floor discussion. This was then followed by three sets of breakout sessions covering:

1. Questions on general improvements of HIA methods with groups given three sets of questions: (i) ‘How to make HIA more accessible and user/practitioner friendly?’ (ii) ‘How do we improve the reliability of HIA?’ and (iii) ‘How to structure HIA in the landscape?’
2. A question on ‘What indicators should HIA focus on for these habitats?’, which was repeated for three habitat types: (i) blanket bog; (ii) dwarf shrub heath, and (iii) tussock grassland. Participants were then asked to vote for the top three indicators identified collectively for each habitat type.
3. A discussion on the development, wording and interpretability of impact classes of the top three indicators identified in the question above for each habitat type. 

There were three groups of approximately five participants in each breakout group. All breakout groups were facilitated by a member of the JHI project team who transcribed discussion in the group verbatim onto a flip chart. Prior to the second breakout session participants were presented with results from the earlier desk-based cross tabulation of indicators and the online stakeholder survey. The first two breakout sessions had a carousel format (City St George's University of London, 2025). A group spent the longest time on the first question, or habitat, and moved around to other questions and habitats, contributing to input from the previous groups for a shorter time. After the second breakout session participants were given three sticky dots for each habitat and asked to vote for their choice of top indicators for that habitat. The workshop ended with an open floor discussion on HIA methods that was transcribed verbatim by the project team note takers. All flipcharts and voting scores of indicators were transcribed after the event and presented in the results. 

Results

Cross-tabulation of HIA indicators

A total of 76 indicators were identified across the five HIA methods and four habitat types assessed. The MacDonald et al. method had on average three times more indicators than other HIA methods, noting that the pre-MacDonald method did not have tussock grassland indicators (Figure 1a). Across habitat types, blanket bog had the highest number of distinct indicators (27 indicators), followed by dwarf shrub (25 indicators), tussock grassland (22 indicators) and finally smooth grassland (18 indicators). 

In total and across habitat types 44 indicators were not shared and used in only one HIA method (Figure 1b, c). The MacDonald et al. method had far the greater number of non-shared indicators totalling 31 indicators compared to 5 non-shared indicators for WDBP, 3 indicators for CSM and pre-MacDonald and 2 indicators for the Putman method (Figure 2b). Smooth grassland had most shared indicators with only 3 indicators non-shared between HIA methods, whereas dwarf shrub and tussock grassland had 15 non-shared indicators and blanket bog had 11 non-shared indicators (Figure 2b, c). 

By each habitat type, there were three indicators, four indicators, 17 indicators and 24 indicators shared across all five, four, three and two HIA methods, respectively (Figure 1c). Distinct indicators shared across more than four HIA methods were as follows: 

• Browsing of heather (Calluna vulgaris) and/or blaeberry (Vaccinium myrtillus) [blanket bog]
• Signs of grazing on leaves of palatable species (collectively) Agrostis canina, Festuca ovina, and F. vivipara [smooth grassland]
• Sward height and texture [smooth grassland]
• Amount of herbivore dung [dwarf shrub heath]
• Heather stem breakage due to large herbivore trampling [dwarf shrub heath]
• Amount of bare ground [smooth grassland, tussock grassland] 

It is worth noting that HIA methods could share indicators however impact class descriptions within indicators could differ for a shared indicator between HIA methods. Due to the greater consistency of indicators across HIA methods for smooth grassland, these were not assessed by stakeholders using the online survey or stakeholder workshop, which also helped simplify and reduce time required for the online survey and in-person workshop. 

Online stakeholder survey 

In total, 39 respondents answered the survey and of these 30 respondents provided answers to indicator questions. Occupations of those answering the survey encapsulated the diverse user range of HIA users, including: 30% estate manager/estate owner/deer stalker (9 respondents), 27% consultants (8 respondents), 23% government organisations/agencies (7 respondents), 20% researchers (6 respondents) and the remaining respondents identified as a mixture of these professions. Field experience of HIA users ranged from 2 to 30 years with a mean 12.5 years and median 10 years HIA user experience. 

Respondents were asked to select up to two HIA methods they were most familiar with using, thus in total the survey had 40 sets of answers for indicators across different HIA methods. Respondents were most familiar with WDBP comprising 55% (22 responses), followed by MacDonald et al. method 38% (15 responses), and Putman method 7% (3 responses). There were no responses for the pre-MacDonald method. 

We found little differences between HIA methods in whether stakeholders recorded indicators frequently or rarely, viewed them as useful or informative or viewed them as easily confused with non-herbivore impacts (Figure 2). The only difference was indicators were selected more frequently for the Putman method compared to MacDonald method (Figure 2a, Table 1); however, the Putman method only had three responses. 

Comparing habitat types, indicators for tussock grasslands were viewed as less useful and informative and selected less frequently than other habitat types (Figure 2a & c; Table 1). It is worth noting that tussock grasslands in all HIA methods represented a range of tussock grassland types, for instance tussock grassland dominated by purple moor grass (Molinia caerulea), mat grass (Nardus stricta) and tufted hair grass (Deschampsia cespitosa) as such indicators may not be universally relevant, e.g., indicators with species present that are not relevant to all tussock grasslands. Further, the WDBP tussock grassland indicators presented in the survey has not been published by NatureScot, so respondents that were familiar with WDBP, may not have been familiar with the WDBP tussock grassland indicators. Dwarf shrub heath indicators were recorded more frequently than the other two habitat types (Figure 2a; Table 1). There were no significant differences across habitat types in whether indicators were viewed as easily confused with non-herbivore impacts (Table 1). 

Overall, across HIA methods and habitat types, existing indicators average 52 ± 13 % (mean ± standard deviation, median 56%) usefulness and informativeness score highlighting significant scope for the improvement and refinement of indicators. The strongest difference was by habitat type, with indicators for tussock grasslands having the lowest usefulness score averaging 39% (median 44%) compared to similar dwarf shrub heath at 58% (median 59%) and blanket bog 60% (median 57%) usefulness scores. Thus, the survey identified tussock grasslands as having the fewest number of useful indicators and thus the strongest need for improvement. 

There were strong correlations between the different evaluation categories assigned to indicators by stakeholders (Figure 3; Table 2). If stakeholders viewed an indicator as frequently recorded, then the same indicator was viewed as useful and informative as evident by a significant and positive correlation between the proportion of stakeholders assigning frequency of use and usefulness and informative to an indicator (Figure 3a: Table 2). Unsurprisingly, indicators that were frequently recorded were also not recorded rarely (Figure 3b; Table 2) and rarely recorded indicators were viewed as less useful and informative to assess herbivore impacts (Figure 3c; Table 2). There was no significant correlation for indicators being viewed as easily confused with non-herbivore impacts with any other evaluation category (Figure 3d; Table 2; other correlations not shown). If an indicator was viewed as useful/informative or frequently recorded it was significantly and positively correlated with the number of times that indicator was shared across HIA methods (see above cross-tabulation of indicators) (Figure 3e useful / informative correlation only; Table 2). Therefore, the most useful and informative (and most frequently used) indicators were the ones most likely to be found in several, or all, of the existing HIA methods. 

Stakeholder workshop 

The general points resulting from the stakeholder workshop were: 

• There was agreement between stakeholders that HIA methods need to be robust and repeatable, whilst at the same time methodologies that will be widely used by practitioners are sufficiently simple for wide stakeholder uptake. However, several stakeholders thought these two aims could not be achieved in single HIA method for both professionals and practitioners.
• Amongst some stakeholders there was an openness to revising HIA methods, for instance, an acknowledgement that the WDBP method had not been revised for many years and this needed addressing.
• Although there was a willingness for some revisions, there were concerns about changing methods as these would not be compatible with previous surveys to assess change.
• Several participants noted a lack of rigorous testing of all HIA methods aside from some work on surveyor consistency. If a HIA method is required to inform deer management and enforce deer regulation greater testing would strengthen the validity of the approach, for example in-depth field testing with quantitative herbivore utilisation approaches; although stakeholders raised that the latter is time consuming and expensive.
• For HIA methods to be adopted by estate managers and deer stalkers requires ‘buy in’, a shared understanding of the objective of the HIA, use of language relevant to practitioners (i.e., use terms like impact rather than damage), and a sense of connection to data collection and analysis such as smart phone applications. Stakeholders raised the issue of a current mismatch in the language between proposed revised deer management legislation using terms such as deer damage compared to the assessment terminology of impact. Participants preferred retaining the current terminology of ‘impact’ as this is less subjective.
• There was no consensus on a sampling approach comparing fixed/non-fixed, marked/unmarked. It was noted that if stratifying by habitat, practitioners need GIS skills/local knowledge.
• There was a discussion around the idea of ‘habitat fluidity’ of having a HIA method without grouping by habitat type with the reason being open hill range vegetation is in a constant state of transition, for example between heathlands to grassland or dwarf shrub to woodland, with several states in between. Another argument for a non-habitat orientated HIA method was that deer stalkers and estate managers use the geography of deer movement (i.e., summer/winter hefting ground) rather than habitat types per se. There was disagreement on this topic of habitat fluidity because habitat types operationally give a defined plant community that differs in potential herbivore preference and impact, and dominant vegetation types provide a way to spatially scale up recorded impacts. Whilst a habitat-fluid method would be interesting to test, the consensus was to keep structuring HIAs around habitat type.
• Stakeholders noted that the Woodland HIA has been regularly updated over two decades in contrast to the open hill rangeland HIA, and this has been strongly driven by ecologist Helen Armstrong and funded by NatureScot, Forest Land-use Scotland and Scottish Forestry. This has allowed stronger stakeholder consensus on methodical approaches. In the Woodland HIA indicators have been refined based upon expert knowledge and interpretability of indicators refined through  surveyor consistency tests (Armstrong, 2021). Although, like the open hill range HIA, there are no scientific publications supporting the Woodland HIA, and some stakeholders thought linking indicators / impact classes with detailed measures of herbivore impact would give stronger grounding for thresholds and the future prognosis of habitat condition following herbivory.   
• Areas suggested for future development of open hill range HIA included: (1) skills and training to improve the reliability of HIA outcomes from designing surveys to data handling and analysis.  Earlier funding for HIA training by NatureScot to DMGs had widespread uptake of over 40 DMGs (NatureScot, 2019); (2) regular updates to the open hill rangeland method; (3) a central accessible database for HIA data, although stakeholders did note a previous initiative to compile a deer management database led by NatureScot (formerly SNH) through Supporting Wide Area Range Management for Deer (SWARD) (NatureScot, 2019); (4) testing of different methods and surveyor variation, and (5) development of application for smart phones / tablets, to engage deer stalkers and estate managers, including aspects of sampling design, data input and processing, and data interpretation i.e., would give the impact class based on data input.

General principles of revising indicators and harmonising indicators were identified: 

Accessible language for practitioners
Clear description of indicators to reduce variation in surveyor interpretation
Widen the number of impact classes (sensu MacDonald et al. method addendum (MacDonald, 2007) and Woodland HIA (Armstrong et al., 2023)) 
Inclusion of an impact class that is described as ‘no impact’ or ‘negligible impact’ 
Highlight scientific underpinning of indicators, particularly for the impact of herbivores in relation to sustaining a habitat type, habitat feature or specific plant species
Consideration of differential herbivore impacts on herbivore preferred and unpreferred species (stakeholders used terms of palatable and unpalatable species)
Harmonisation of indicators with the Woodland Herbivore Impact Assessment
Desire for more quantitative indicators, but keep measurement approach simple

Workshop indicators by habitat type 

As part of the workshop, stakeholders discussed indicators and impact classes relating to three habitat types: blanket bogs, dwarf shrub heath and tussock grassland. Within each habitat type, stakeholders voted on the most useful/informative indicator and further described impact classes within those indicators. The exercise raised 14 alternative indicators not already existing in HIA methods for those habitats and gave insights into the required description indicators and impact classes. Below we list the indicators identified for each habitat type and notes on impact classes for indicators and votes in parenthesis. 

Blanket Bog

Indicators

1. Trampling / hoof prints in bare peat and vegetation and severity and extent of tracking by herbivores (12 votes)
2. Assessment of browsing patterns and severity of browsing on Calluna vulgaris, Vaccinium myrtillus and Myrica gale bushes should be able to assess both current and longer-term trends in herbivore impacts. There was debate about excluding V. myrtillus as an indicator as it is not a common component of Scottish blanket bogs (10 votes)
3. The grazing / browsing levels on less palatable species can be useful where impacts are relatively high – a palatability index of the most common plants in such habitats would be useful (9 votes)
4. Extent and severity of trampling damage specifically on Sphagnum lawns and hummocks (8 votes)
5. Dung counts in walked transects or plots can be relatively quick and useful data to collect to compare herbivore presence over time and identify the type of herbivore presence. Over the long-term higher dung counts could have ecological consequences, such as soil nutrient enrichment (1 vote)

Impact class 

1. Indicator: Trampling/hoof prints in bare peat and vegetation and severity and extent of tracking by herbivores.
   Impact class: Add an Absent / No Impacts category, e.g. if areas fenced off or exceedingly low impacts are present. The other three levels of impacts described in MacDonald are OK as they are.
2. Indicator: Assessment of browsing patterns and severity of browsing on Calluna vulgaris and Vaccinium myrtillus bushes 
   Impact class: Could be changed from two classes in MacDonald to three classes.
3. Indicator: The grazing/browsing levels on less palatable species can be useful where impacts are relatively high. 
   Impact class: Maintain the two classes of MacDonald here but consider adding a palatability index of the most common plants to help the objectivity of assessing impact levels.

Statistical considerations

An individual survey at one point in time reveals important information about current levels of herbivore impact. However, much more information can be gained by analysing changes in impact through time.

Largely, analysis has focussed on the changing number of points within habitats that are classified in the different impact categories. Alternative approaches have turned the categories into a numeric scale, but this makes the considerable assumption that the scale is linear.

Herbivore Impact Assessments (HIA) generate ordinal data, categorical variables that have an order or hierarchy, i.e., from low to high herbivore impact. There are statistical methods that have been developed for ordinal data, so that changes through time can be assessed to see if they are robust. This is likely not necessary for assessing change through time for an individual land holding, but it would be best practice to use appropriate statistical methods for large scale comparisons. It is also useful to be able to assess change using data on individual indicators as well as on overall impact classes to better identify what is driving change.

Formal statistical analysis of ordinal data is not straightforward and may not be necessary at the level of a land holding. However, it would be useful to create guidelines for how to interpret data, in effect, providing guidance on whether a level of change for a given sampling intensity is likely to reflect real change or statistical noise. Larger-scale analyses would require more formal testing.

Stakeholder feedback

On 5^th April 2025, the draft report, user guide and technical annex was shared with stakeholders who attended the workshop and Scientific Advisory Committee Deer Subgroup for NatureScot for feedback. A total of ten stakeholders reviewed the documents and provided feedback. We thank all stakeholders for their feedback. Feedback was used to revise and improve the body of text, nevertheless we summarise major concerns raised below: 

• Does HIA need revising? Some stakeholders asked the question what evidence is there that existing HIA methods do not work? Stakeholders from the ecological consultancy community highlighted a lack of transparency from SNH / NatureScot and JHI of a comprehensive analysis showing issues with existing HIA methods. Going further, some stakeholders requested HIA should be assessed by an academic institution outside of Scotland to ensure full independence of any findings. In response to this point, NatureScot raised the tender for this project due to inconsistencies in HIA outcomes from using different HIA methods. Data analysis of inconsistencies in HIA methods was not part of this tender. We have recommended to NatureScot that issues of these inconsistencies due to HIA methods should be analysed and published to highlight the problem and ensure stakeholder support for revisions to existing HIA methods.
• A stronger distinction between what HIA methods achieve: some stakeholders stressed the point that HIA assesses the level of grazing / trampling, and not the impact of herbivores on the species composition or structure of the vegetation. The critique of this report was that a failure to understand this difference undermined the rigour in the whole approach to this study. We do not agree with this criticism as different indicators used in HIA method attempt to assess either or both grazing / trampling levels and vegetation impact.
• Differences between HIA user groups: some professional consultants raised concerns that they did not think it possible to obtain both a robust method for assessing herbivore impact and one that is easily accessible and used by stalkers and estate managements from the same method. Their justification being highly experienced ecologist are required to take account of many indicators and the environmental context to make an overall assessment of the levels of grazing / browsing or trampling and how these may result in change of the vegetation. Instead, practitioners have more experience and understanding of the behaviour of large animals, but usually less knowledge of vegetation dynamics and ecosystem processes. Several professional consultants believe a two-tiered approach would be required as a single method may not be adequate for all parties (consultants, estate managers, deer stalkers) surveying. Others raised issues more generally that different user groups have different agendas, and as a result produce consistently different results to suit respective aims. In the current form, the two-tiered HIA methods is contributing to inconsistent HIA outcomes. At present, we have devised a single method, and leaned more towards simplifying the method, this could be split in two, but with the need to ensure strong alignment between the assessments provided by the different HIA methods.   
• Reductionist approach and dumbing down will reduce robustness: stakeholders raised concern on the focus of reducing indicators. They noted loss of potentially rare but useful indicators have been removed and note the value of species-specific indicators or grouping by palatability. Field testing could incorporate a wider suite of indicators found in Table 4 and compare this wider suite to a reduced subset.
• Spatial methods - walk-through, permanent plots and MLURI rapid assessment method: our recommendation of using fixed-point plot-based was strongly disagreed with by some stakeholders who would have preferred a walk-through method. Whilst a random walk-through method likely produces similar results to fixed-point sampling, there is the issue that the randomness brought in during the walk-through method reduces the power available to detect change compared to revisiting fixed points (Barker, 2001). We agree with the point of potential loss of time looking for marked fixed points and thus suggested the use of GPS devices to relocate previously sampled points. Some stakeholders raised the issue that the MLURI rapid assessment method was not shared with stakeholders, and they would like further consultation on this design.
• Terminology and use of the term ‘damage’: stakeholders said it was unclear from the report the role of HIA in NatureScot's regulatory decision making. Previously HIA assessments have been used to judge whether Section 7 agreements were working. However, on NatureScot’s website the terminology is stated as ‘damage’ to the Natural Heritage. Stakeholders point out that there seems to be no read-across of the use of ‘damage’ in proposed changes to HIA, and thus what then is the purpose of HIA? Is it the case that the workshop should have been focused on discussing Site Condition Monitoring and not HIA? We have outlined at the start of the report that NatureScot uses HIA in regulatory decision making. We have also highlighted as general feedback from the workshop that wording is important and current wording in legislation of ‘damage’ will be problematic for stakeholders and does not have read across of ‘impact’ used in HIA.
• ‘Danger’ of relying too heavily on measurements on heather (Calluna vulgaris) as an indicator. ‘Over-reliance’ on heather alone can be very misleading since browsing impact is heavily frequency-dependent. If there is a blanket of heather across a wide area, impacts of a given density of deer are diluted. If heather is sparse within the sward and represented only by a few stray stems here and there, then (whatever the density of deer) it will always return an apparently high impact because it is preferentially selected. We are aware of such heather-abundance deer-impact effects and highlighted them as consideration in field-testing.
• Tussock grasslands focused on purple moor grass (Molinia caerulea): several stakeholders raised concern on the focus on purple moor tussock grasslands on two grounds: (1) herbivore value and (2) omission of other tussock grasslands. On herbivore value, stakeholders viewed purple moor grass-dominated grasslands as a form of degraded wet heath or degraded blanket bog, thus the existence of this habitat is evidence itself of long-term high herbivore impact and of limited survey value and a better approach would be to look at historic increases in this habitat land cover. On the second, mat grass (Nardus stricta) tussock grasslands are widespread in south of Scotland, these grasslands can be an important grassland in protected areas and a habitat that may show quicker response to herbivore reduction/removal compared to purple moor grass-dominated grassland. The justification for our focus on purple moor grass-dominated tussock grassland is that it covers an extensive land area across the west and north of Scotland and currently is not covered in WDBP leading to it being grouped into either dwarf shrub heath or blanket bog and so contributing to inconsistencies in HIA outcomes. Further, the grouping of indicators for different tussock grasslands in MacDonald method are not universally shared due to different types of tussock grassland, which has in part reduced the usefulness of many of these indicators. If other tussock grasslands are of interest to HIA we suggest habitat-specific development of indicators, for instance for Nardus strica dominated tussock grasslands.
• Mixing of current and historic indicators: some stakeholders raised concern that selected indicators mix current and past / long-term herbivore impacts. These will need to be distinguished and potentially analysed separately as they provide different information about herbivore impacts. 

NatureScot Scientific Advisory Committee (SAC) Deer Subgroup feedback

Feedback from the SAC Deer Subgroup was generally positive, particularly achieving the project in a short duration. Generally, feedback was the following: 

• Questions around the processing and analysis of indicators, for example are indicators equally weighted? what to do with data? what ordinal analysis to apply? This was not part of the remit of this tender, but it will be integral to the next phase of field-testing.
• Aggregation versus separation of indicators: the committee recommended keeping indicators separate when analysing and minimising aggregation, because individual indicators could provide different sources of information on herbivore impact. This was based on committee members experience with reviewing indicator methodologies for windfarm impacts on seabirds.
• Improve the user-guide interface so that it could be clearly interpreted by a new user rather than someone familiar with the report and details in the annex. It was agreed further refinement of user-guide could wait until after field-testing.

Recommendations

We make the following recommendations for next steps in this process:

• There should be field-testing using the indicators suggested. Given that many of the indicators have thresholds that are not based on clear ecological evidence, such a test would include noting the raw data for each indicator to better revise the thresholds for each indicator. For instance, for browsing of heather, instead of recording which category browsing fell into at a point, recording the estimated level of browsing would allow for better comparison of thresholds between indicators. Correlation analysis of raw data would also identify indicators that were not behaving in the same way as other indicators.
• Field testing should also consider the wider breadth of indicators selected in this user guide. Several indicators have been dropped due to rarity of use but should still be considered in field testing. Stakeholders raised a number of issues around splitting or amending some indicators and split and/or amended versions should also be investigated during field testing.
• Field-testing could assess the consistency of recorders. Assuming that recorders see the same thing, then inconsistency would indicate different interpretations of the thresholds for the indicator. The language used in the descriptions could then be improved.
• Field-testing should include a range of likely users, from NatureScot staff, deer stalkers/estate managers and consultants, so that views on the operation of the consolidated methods is gathered from a diverse group of users.
• Desk-based testing is required to investigate whether sampling designs at different spatial scales can be amalgamated and produce consistent results, namely investigating sample selection of walk-through versus fixed point sampling designs and comparing sampling designs for smaller and larger estates.
• Field-testing should also consider user-friendly approaches to data recording and processing. For example, standardised templates for recording and excel templates with supporting formulae that allow easy analysis of collected data. Field-testing should integrate both assigned categories and raw data in analysis.
• On circulation of the report, user guide, and technical annex to the stakeholder community, several concerns were raised which we suggest NatureScot take on board. One of importance would be analysis and publication of the proportion or extent of inconsistencies of HIA outcomes from different HIA methods that formed the basis of this tender. Publication of such information would allow transparency of the issues underlying existing HIA and the need for revisions.
• After field-testing and a period of bedding-in of a refined HIA method several stakeholders expressed an ambition for the HIA method to be translated into a smartphone and tablet application. An App would support uptake by a diverse range of practitioners and would benefit deer stalkers and estate managers that collect, analyse and view data in real time and compare to previous survey results. We stress at the open-hill range HIA needs field testing and refinement, but an app would be an ambition to keep in mind for later development.

User guide 

Introduction

The purpose of this Herbivore Impact Assessment (HIA) is to determine the impact of herbivores on open-hill ranges’ vegetation using features that have changed as a result of herbivore presence. The main herbivores of interest are deer, sheep, cattle and to a lesser extent smaller herbivores such as hares, rabbits and voles. As part of the HIA a surveyor records indicators, including evidence of grazing or browsing of vegetation, changes in plant growth-form and species composition, evidence of disturbance and trampling. For each indicator there are different severities of herbivore impact which are then assigned an impact class. This indicator method builds on previous HIA methods, including MacDonald et al. method (also referred previously as the Scottish Natural Heritage method) (MacDonald et al., 1998a, MacDonald et al., 1998b, MacDonald, 2007); Wild Deer Best Practice (WDBP) (Wild Deer Best Practice Guidance for Scotland, 2018), and the Putman method. The method has been designed for open hill rangeland habitats in Scotland, specifically blanket bogs, dwarf shrub heaths, tussock grasslands and smooth grasslands. 

In support of this user guide, there is a technical annex that provides reasoning for the spatial design, habitat description and indicators selected. 

Sampling design

For an entire estate or within-estate survey, it is recommended that the selection of sampling points be based on a grid intersect method. This sampling approach reduces bias (e.g., bias of sampling through accessibility along roads or paths) and ensures random selection of points when background knowledge of habitat type distribution is unknown. This can be done by overlaying an Ordnance Survey map or similar with a uniformly spaced grid to select sampling points at each grid intersect. The minimum number of survey points should ideally be above 30 – 50 points per habitat type to be statistically robust, but the exact number and density will depend on the area to be surveyed and cost and labour time. Software such GIS/qGIS software can be used to look at OS maps, an accompanying aerial photograph and overlaying grid lines. There are alternatives to GIS software which are freely available online such as Where's The Path 3 (example survey designs in Figure 4). 

Statistical considerations

An individual survey at one point in time reveals important information about current levels of herbivore impact. However, much more information can be gained by analysing changes in impact through time.

Largely, analysis has focussed on the changing number of points within habitats that are classified in the different impact categories. Alternative approaches have turned the categories into a numeric scale, but this makes the considerable assumption that the scale is linear.

Herbivore Impact Assessments (HIA) generate ordinal data, categorical variables that have an order or hierarchy, i.e., from low to high herbivore impact. There are statistical methods that have been developed for ordinal data, so that changes through time can be assessed to see if they are robust. This is likely not necessary for assessing change through time for an individual land holding, but it would be best practice to use appropriate statistical methods for large scale comparisons. It is also useful to be able to assess change using data on individual indicators as well as on overall impact classes to better identify what is driving change.

Formal statistical analysis of ordinal data is not straightforward and may not be necessary at the level of a land holding. However, it would be useful to create guidelines for how to interpret data, in effect, providing guidance on whether a level of change for a given sampling intensity is likely to reflect real change or statistical noise. Larger-scale analyses would require more formal testing.

Stakeholder feedback

On 5^th April 2025, the draft report, user guide and technical annex was shared with stakeholders who attended the workshop and Scientific Advisory Committee Deer Subgroup for NatureScot for feedback. A total of ten stakeholders reviewed the documents and provided feedback. We thank all stakeholders for their feedback. Feedback was used to revise and improve the body of text, nevertheless we summarise major concerns raised below: 

• Does HIA need revising? Some stakeholders asked the question what evidence is there that existing HIA methods do not work? Stakeholders from the ecological consultancy community highlighted a lack of transparency from SNH / NatureScot and JHI of a comprehensive analysis showing issues with existing HIA methods. Going further, some stakeholders requested HIA should be assessed by an academic institution outside of Scotland to ensure full independence of any findings. In response to this point, NatureScot raised the tender for this project due to inconsistencies in HIA outcomes from using different HIA methods. Data analysis of inconsistencies in HIA methods was not part of this tender. We have recommended to NatureScot that issues of these inconsistencies due to HIA methods should be analysed and published to highlight the problem and ensure stakeholder support for revisions to existing HIA methods.
• A stronger distinction between what HIA methods achieve: some stakeholders stressed the point that HIA assesses the level of grazing / trampling, and not the impact of herbivores on the species composition or structure of the vegetation. The critique of this report was that a failure to understand this difference undermined the rigour in the whole approach to this study. We do not agree with this criticism as different indicators used in HIA method attempt to assess either or both grazing / trampling levels and vegetation impact.
• Differences between HIA user groups: some professional consultants raised concerns that they did not think it possible to obtain both a robust method for assessing herbivore impact and one that is easily accessible and used by stalkers and estate managements from the same method. Their justification being highly experienced ecologist are required to take account of many indicators and the environmental context to make an overall assessment of the levels of grazing / browsing or trampling and how these may result in change of the vegetation. Instead, practitioners have more experience and understanding of the behaviour of large animals, but usually less knowledge of vegetation dynamics and ecosystem processes. Several professional consultants believe a two-tiered approach would be required as a single method may not be adequate for all parties (consultants, estate managers, deer stalkers) surveying. Others raised issues more generally that different user groups have different agendas, and as a result produce consistently different results to suit respective aims. In the current form, the two-tiered HIA methods is contributing to inconsistent HIA outcomes. At present, we have devised a single method, and leaned more towards simplifying the method, this could be split in two, but with the need to ensure strong alignment between the assessments provided by the different HIA methods.   
• Reductionist approach and dumbing down will reduce robustness: stakeholders raised concern on the focus of reducing indicators. They noted loss of potentially rare but useful indicators have been removed and note the value of species-specific indicators or grouping by palatability. Field testing could incorporate a wider suite of indicators found in Table 4 and compare this wider suite to a reduced subset.
• Spatial methods - walk-through, permanent plots and MLURI rapid assessment method: our recommendation of using fixed-point plot-based was strongly disagreed with by some stakeholders who would have preferred a walk-through method. Whilst a random walk-through method likely produces similar results to fixed-point sampling, there is the issue that the randomness brought in during the walk-through method reduces the power available to detect change compared to revisiting fixed points (Barker, 2001). We agree with the point of potential loss of time looking for marked fixed points and thus suggested the use of GPS devices to relocate previously sampled points. Some stakeholders raised the issue that the MLURI rapid assessment method was not shared with stakeholders, and they would like further consultation on this design.
• Terminology and use of the term ‘damage’: stakeholders said it was unclear from the report the role of HIA in NatureScot's regulatory decision making. Previously HIA assessments have been used to judge whether Section 7 agreements were working. However, on NatureScot’s website the terminology is stated as ‘damage’ to the Natural Heritage. Stakeholders point out that there seems to be no read-across of the use of ‘damage’ in proposed changes to HIA, and thus what then is the purpose of HIA? Is it the case that the workshop should have been focused on discussing Site Condition Monitoring and not HIA? We have outlined at the start of the report that NatureScot uses HIA in regulatory decision making. We have also highlighted as general feedback from the workshop that wording is important and current wording in legislation of ‘damage’ will be problematic for stakeholders and does not have read across of ‘impact’ used in HIA.
• ‘Danger’ of relying too heavily on measurements on heather (Calluna vulgaris) as an indicator. ‘Over-reliance’ on heather alone can be very misleading since browsing impact is heavily frequency-dependent. If there is a blanket of heather across a wide area, impacts of a given density of deer are diluted. If heather is sparse within the sward and represented only by a few stray stems here and there, then (whatever the density of deer) it will always return an apparently high impact because it is preferentially selected. We are aware of such heather-abundance deer-impact effects and highlighted them as consideration in field-testing.
• Tussock grasslands focused on purple moor grass (Molinia caerulea): several stakeholders raised concern on the focus on purple moor tussock grasslands on two grounds: (1) herbivore value and (2) omission of other tussock grasslands. On herbivore value, stakeholders viewed purple moor grass-dominated grasslands as a form of degraded wet heath or degraded blanket bog, thus the existence of this habitat is evidence itself of long-term high herbivore impact and of limited survey value and a better approach would be to look at historic increases in this habitat land cover. On the second, mat grass (Nardus stricta) tussock grasslands are widespread in south of Scotland, these grasslands can be an important grassland in protected areas and a habitat that may show quicker response to herbivore reduction/removal compared to purple moor grass-dominated grassland. The justification for our focus on purple moor grass-dominated tussock grassland is that it covers an extensive land area across the west and north of Scotland and currently is not covered in WDBP leading to it being grouped into either dwarf shrub heath or blanket bog and so contributing to inconsistencies in HIA outcomes. Further, the grouping of indicators for different tussock grasslands in MacDonald method are not universally shared due to different types of tussock grassland, which has in part reduced the usefulness of many of these indicators. If other tussock grasslands are of interest to HIA we suggest habitat-specific development of indicators, for instance for Nardus strica dominated tussock grasslands.
• Mixing of current and historic indicators: some stakeholders raised concern that selected indicators mix current and past / long-term herbivore impacts. These will need to be distinguished and potentially analysed separately as they provide different information about herbivore impacts. 

NatureScot Scientific Advisory Committee (SAC) Deer Subgroup feedback

Feedback from the SAC Deer Subgroup was generally positive, particularly achieving the project in a short duration. Generally, feedback was the following: 

• Questions around the processing and analysis of indicators, for example are indicators equally weighted? what to do with data? what ordinal analysis to apply? This was not part of the remit of this tender, but it will be integral to the next phase of field-testing.
• Aggregation versus separation of indicators: the committee recommended keeping indicators separate when analysing and minimising aggregation, because individual indicators could provide different sources of information on herbivore impact. This was based on committee members experience with reviewing indicator methodologies for windfarm impacts on seabirds.
• Improve the user-guide interface so that it could be clearly interpreted by a new user rather than someone familiar with the report and details in the annex. It was agreed further refinement of user-guide could wait until after field-testing.

Recommendations

We make the following recommendations for next steps in this process:

• There should be field-testing using the indicators suggested. Given that many of the indicators have thresholds that are not based on clear ecological evidence, such a test would include noting the raw data for each indicator to better revise the thresholds for each indicator. For instance, for browsing of heather, instead of recording which category browsing fell into at a point, recording the estimated level of browsing would allow for better comparison of thresholds between indicators. Correlation analysis of raw data would also identify indicators that were not behaving in the same way as other indicators.
• Field testing should also consider the wider breadth of indicators selected in this user guide. Several indicators have been dropped due to rarity of use but should still be considered in field testing. Stakeholders raised a number of issues around splitting or amending some indicators and split and/or amended versions should also be investigated during field testing.
• Field-testing could assess the consistency of recorders. Assuming that recorders see the same thing, then inconsistency would indicate different interpretations of the thresholds for the indicator. The language used in the descriptions could then be improved.
• Field-testing should include a range of likely users, from NatureScot staff, deer stalkers/estate managers and consultants, so that views on the operation of the consolidated methods is gathered from a diverse group of users.
• Desk-based testing is required to investigate whether sampling designs at different spatial scales can be amalgamated and produce consistent results, namely investigating sample selection of walk-through versus fixed point sampling designs and comparing sampling designs for smaller and larger estates.
• Field-testing should also consider user-friendly approaches to data recording and processing. For example, standardised templates for recording and excel templates with supporting formulae that allow easy analysis of collected data. Field-testing should integrate both assigned categories and raw data in analysis.
• On circulation of the report, user guide, and technical annex to the stakeholder community, several concerns were raised which we suggest NatureScot take on board. One of importance would be analysis and publication of the proportion or extent of inconsistencies of HIA outcomes from different HIA methods that formed the basis of this tender. Publication of such information would allow transparency of the issues underlying existing HIA and the need for revisions.
• After field-testing and a period of bedding-in of a refined HIA method several stakeholders expressed an ambition for the HIA method to be translated into a smartphone and tablet application. An App would support uptake by a diverse range of practitioners and would benefit deer stalkers and estate managers that collect, analyse and view data in real time and compare to previous survey results. We stress at the open-hill range HIA needs field testing and refinement, but an app would be an ambition to keep in mind for later development.

User guide 

Introduction

The purpose of this Herbivore Impact Assessment (HIA) is to determine the impact of herbivores on open-hill ranges’ vegetation using features that have changed as a result of herbivore presence. The main herbivores of interest are deer, sheep, cattle and to a lesser extent smaller herbivores such as hares, rabbits and voles. As part of the HIA a surveyor records indicators, including evidence of grazing or browsing of vegetation, changes in plant growth-form and species composition, evidence of disturbance and trampling. For each indicator there are different severities of herbivore impact which are then assigned an impact class. This indicator method builds on previous HIA methods, including MacDonald et al. method (also referred previously as the Scottish Natural Heritage method) (MacDonald et al., 1998a, MacDonald et al., 1998b, MacDonald, 2007); Wild Deer Best Practice (WDBP) (Wild Deer Best Practice Guidance for Scotland, 2018), and the Putman method. The method has been designed for open hill rangeland habitats in Scotland, specifically blanket bogs, dwarf shrub heaths, tussock grasslands and smooth grasslands. 

In support of this user guide, there is a technical annex that provides reasoning for the spatial design, habitat description and indicators selected. 

Sampling design

For an entire estate or within-estate survey, it is recommended that the selection of sampling points be based on a grid intersect method. This sampling approach reduces bias (e.g., bias of sampling through accessibility along roads or paths) and ensures random selection of points when background knowledge of habitat type distribution is unknown. This can be done by overlaying an Ordnance Survey map or similar with a uniformly spaced grid to select sampling points at each grid intersect. The minimum number of survey points should ideally be above 30 – 50 points per habitat type to be statistically robust, but the exact number and density will depend on the area to be surveyed and cost and labour time. Software such GIS/qGIS software can be used to look at OS maps, an accompanying aerial photograph and overlaying grid lines. There are alternatives to GIS software which are freely available online such as Where's The Path 3 (example survey designs in Figure 4). 

Statistical considerations

An individual survey at one point in time reveals important information about current levels of herbivore impact. However, much more information can be gained by analysing changes in impact through time.

Largely, analysis has focussed on the changing number of points within habitats that are classified in the different impact categories. Alternative approaches have turned the categories into a numeric scale, but this makes the considerable assumption that the scale is linear.

Herbivore Impact Assessments (HIA) generate ordinal data, categorical variables that have an order or hierarchy, i.e., from low to high herbivore impact. There are statistical methods that have been developed for ordinal data, so that changes through time can be assessed to see if they are robust. This is likely not necessary for assessing change through time for an individual land holding, but it would be best practice to use appropriate statistical methods for large scale comparisons. It is also useful to be able to assess change using data on individual indicators as well as on overall impact classes to better identify what is driving change.

Formal statistical analysis of ordinal data is not straightforward and may not be necessary at the level of a land holding. However, it would be useful to create guidelines for how to interpret data, in effect, providing guidance on whether a level of change for a given sampling intensity is likely to reflect real change or statistical noise. Larger-scale analyses would require more formal testing.

Stakeholder feedback

On 5^th April 2025, the draft report, user guide and technical annex was shared with stakeholders who attended the workshop and Scientific Advisory Committee Deer Subgroup for NatureScot for feedback. A total of ten stakeholders reviewed the documents and provided feedback. We thank all stakeholders for their feedback. Feedback was used to revise and improve the body of text, nevertheless we summarise major concerns raised below: 

• Does HIA need revising? Some stakeholders asked the question what evidence is there that existing HIA methods do not work? Stakeholders from the ecological consultancy community highlighted a lack of transparency from SNH / NatureScot and JHI of a comprehensive analysis showing issues with existing HIA methods. Going further, some stakeholders requested HIA should be assessed by an academic institution outside of Scotland to ensure full independence of any findings. In response to this point, NatureScot raised the tender for this project due to inconsistencies in HIA outcomes from using different HIA methods. Data analysis of inconsistencies in HIA methods was not part of this tender. We have recommended to NatureScot that issues of these inconsistencies due to HIA methods should be analysed and published to highlight the problem and ensure stakeholder support for revisions to existing HIA methods.
• A stronger distinction between what HIA methods achieve: some stakeholders stressed the point that HIA assesses the level of grazing / trampling, and not the impact of herbivores on the species composition or structure of the vegetation. The critique of this report was that a failure to understand this difference undermined the rigour in the whole approach to this study. We do not agree with this criticism as different indicators used in HIA method attempt to assess either or both grazing / trampling levels and vegetation impact.
• Differences between HIA user groups: some professional consultants raised concerns that they did not think it possible to obtain both a robust method for assessing herbivore impact and one that is easily accessible and used by stalkers and estate managements from the same method. Their justification being highly experienced ecologist are required to take account of many indicators and the environmental context to make an overall assessment of the levels of grazing / browsing or trampling and how these may result in change of the vegetation. Instead, practitioners have more experience and understanding of the behaviour of large animals, but usually less knowledge of vegetation dynamics and ecosystem processes. Several professional consultants believe a two-tiered approach would be required as a single method may not be adequate for all parties (consultants, estate managers, deer stalkers) surveying. Others raised issues more generally that different user groups have different agendas, and as a result produce consistently different results to suit respective aims. In the current form, the two-tiered HIA methods is contributing to inconsistent HIA outcomes. At present, we have devised a single method, and leaned more towards simplifying the method, this could be split in two, but with the need to ensure strong alignment between the assessments provided by the different HIA methods.   
• Reductionist approach and dumbing down will reduce robustness: stakeholders raised concern on the focus of reducing indicators. They noted loss of potentially rare but useful indicators have been removed and note the value of species-specific indicators or grouping by palatability. Field testing could incorporate a wider suite of indicators found in Table 4 and compare this wider suite to a reduced subset.
• Spatial methods - walk-through, permanent plots and MLURI rapid assessment method: our recommendation of using fixed-point plot-based was strongly disagreed with by some stakeholders who would have preferred a walk-through method. Whilst a random walk-through method likely produces similar results to fixed-point sampling, there is the issue that the randomness brought in during the walk-through method reduces the power available to detect change compared to revisiting fixed points (Barker, 2001). We agree with the point of potential loss of time looking for marked fixed points and thus suggested the use of GPS devices to relocate previously sampled points. Some stakeholders raised the issue that the MLURI rapid assessment method was not shared with stakeholders, and they would like further consultation on this design.
• Terminology and use of the term ‘damage’: stakeholders said it was unclear from the report the role of HIA in NatureScot's regulatory decision making. Previously HIA assessments have been used to judge whether Section 7 agreements were working. However, on NatureScot’s website the terminology is stated as ‘damage’ to the Natural Heritage. Stakeholders point out that there seems to be no read-across of the use of ‘damage’ in proposed changes to HIA, and thus what then is the purpose of HIA? Is it the case that the workshop should have been focused on discussing Site Condition Monitoring and not HIA? We have outlined at the start of the report that NatureScot uses HIA in regulatory decision making. We have also highlighted as general feedback from the workshop that wording is important and current wording in legislation of ‘damage’ will be problematic for stakeholders and does not have read across of ‘impact’ used in HIA.
• ‘Danger’ of relying too heavily on measurements on heather (Calluna vulgaris) as an indicator. ‘Over-reliance’ on heather alone can be very misleading since browsing impact is heavily frequency-dependent. If there is a blanket of heather across a wide area, impacts of a given density of deer are diluted. If heather is sparse within the sward and represented only by a few stray stems here and there, then (whatever the density of deer) it will always return an apparently high impact because it is preferentially selected. We are aware of such heather-abundance deer-impact effects and highlighted them as consideration in field-testing.
• Tussock grasslands focused on purple moor grass (Molinia caerulea): several stakeholders raised concern on the focus on purple moor tussock grasslands on two grounds: (1) herbivore value and (2) omission of other tussock grasslands. On herbivore value, stakeholders viewed purple moor grass-dominated grasslands as a form of degraded wet heath or degraded blanket bog, thus the existence of this habitat is evidence itself of long-term high herbivore impact and of limited survey value and a better approach would be to look at historic increases in this habitat land cover. On the second, mat grass (Nardus stricta) tussock grasslands are widespread in south of Scotland, these grasslands can be an important grassland in protected areas and a habitat that may show quicker response to herbivore reduction/removal compared to purple moor grass-dominated grassland. The justification for our focus on purple moor grass-dominated tussock grassland is that it covers an extensive land area across the west and north of Scotland and currently is not covered in WDBP leading to it being grouped into either dwarf shrub heath or blanket bog and so contributing to inconsistencies in HIA outcomes. Further, the grouping of indicators for different tussock grasslands in MacDonald method are not universally shared due to different types of tussock grassland, which has in part reduced the usefulness of many of these indicators. If other tussock grasslands are of interest to HIA we suggest habitat-specific development of indicators, for instance for Nardus strica dominated tussock grasslands.
• Mixing of current and historic indicators: some stakeholders raised concern that selected indicators mix current and past / long-term herbivore impacts. These will need to be distinguished and potentially analysed separately as they provide different information about herbivore impacts. 

NatureScot Scientific Advisory Committee (SAC) Deer Subgroup feedback

Feedback from the SAC Deer Subgroup was generally positive, particularly achieving the project in a short duration. Generally, feedback was the following: 

• Questions around the processing and analysis of indicators, for example are indicators equally weighted? what to do with data? what ordinal analysis to apply? This was not part of the remit of this tender, but it will be integral to the next phase of field-testing.
• Aggregation versus separation of indicators: the committee recommended keeping indicators separate when analysing and minimising aggregation, because individual indicators could provide different sources of information on herbivore impact. This was based on committee members experience with reviewing indicator methodologies for windfarm impacts on seabirds.
• Improve the user-guide interface so that it could be clearly interpreted by a new user rather than someone familiar with the report and details in the annex. It was agreed further refinement of user-guide could wait until after field-testing.

Recommendations

We make the following recommendations for next steps in this process:

• There should be field-testing using the indicators suggested. Given that many of the indicators have thresholds that are not based on clear ecological evidence, such a test would include noting the raw data for each indicator to better revise the thresholds for each indicator. For instance, for browsing of heather, instead of recording which category browsing fell into at a point, recording the estimated level of browsing would allow for better comparison of thresholds between indicators. Correlation analysis of raw data would also identify indicators that were not behaving in the same way as other indicators.
• Field testing should also consider the wider breadth of indicators selected in this user guide. Several indicators have been dropped due to rarity of use but should still be considered in field testing. Stakeholders raised a number of issues around splitting or amending some indicators and split and/or amended versions should also be investigated during field testing.
• Field-testing could assess the consistency of recorders. Assuming that recorders see the same thing, then inconsistency would indicate different interpretations of the thresholds for the indicator. The language used in the descriptions could then be improved.
• Field-testing should include a range of likely users, from NatureScot staff, deer stalkers/estate managers and consultants, so that views on the operation of the consolidated methods is gathered from a diverse group of users.
• Desk-based testing is required to investigate whether sampling designs at different spatial scales can be amalgamated and produce consistent results, namely investigating sample selection of walk-through versus fixed point sampling designs and comparing sampling designs for smaller and larger estates.
• Field-testing should also consider user-friendly approaches to data recording and processing. For example, standardised templates for recording and excel templates with supporting formulae that allow easy analysis of collected data. Field-testing should integrate both assigned categories and raw data in analysis.
• On circulation of the report, user guide, and technical annex to the stakeholder community, several concerns were raised which we suggest NatureScot take on board. One of importance would be analysis and publication of the proportion or extent of inconsistencies of HIA outcomes from different HIA methods that formed the basis of this tender. Publication of such information would allow transparency of the issues underlying existing HIA and the need for revisions.
• After field-testing and a period of bedding-in of a refined HIA method several stakeholders expressed an ambition for the HIA method to be translated into a smartphone and tablet application. An App would support uptake by a diverse range of practitioners and would benefit deer stalkers and estate managers that collect, analyse and view data in real time and compare to previous survey results. We stress at the open-hill range HIA needs field testing and refinement, but an app would be an ambition to keep in mind for later development.

User guide 

Introduction

The purpose of this Herbivore Impact Assessment (HIA) is to determine the impact of herbivores on open-hill ranges’ vegetation using features that have changed as a result of herbivore presence. The main herbivores of interest are deer, sheep, cattle and to a lesser extent smaller herbivores such as hares, rabbits and voles. As part of the HIA a surveyor records indicators, including evidence of grazing or browsing of vegetation, changes in plant growth-form and species composition, evidence of disturbance and trampling. For each indicator there are different severities of herbivore impact which are then assigned an impact class. This indicator method builds on previous HIA methods, including MacDonald et al. method (also referred previously as the Scottish Natural Heritage method) (MacDonald et al., 1998a, MacDonald et al., 1998b, MacDonald, 2007); Wild Deer Best Practice (WDBP) (Wild Deer Best Practice Guidance for Scotland, 2018), and the Putman method. The method has been designed for open hill rangeland habitats in Scotland, specifically blanket bogs, dwarf shrub heaths, tussock grasslands and smooth grasslands. 

In support of this user guide, there is a technical annex that provides reasoning for the spatial design, habitat description and indicators selected. 

Sampling design

For an entire estate or within-estate survey, it is recommended that the selection of sampling points be based on a grid intersect method. This sampling approach reduces bias (e.g., bias of sampling through accessibility along roads or paths) and ensures random selection of points when background knowledge of habitat type distribution is unknown. This can be done by overlaying an Ordnance Survey map or similar with a uniformly spaced grid to select sampling points at each grid intersect. The minimum number of survey points should ideally be above 30 – 50 points per habitat type to be statistically robust, but the exact number and density will depend on the area to be surveyed and cost and labour time. Software such GIS/qGIS software can be used to look at OS maps, an accompanying aerial photograph and overlaying grid lines. There are alternatives to GIS software which are freely available online such as Where's The Path 3 (example survey designs in Figure 4). 

Once a suitable number of grid points at an appropriate scale are selected, they can be listed into a survey recording form and/or downloaded into a GPS device as waypoints. The surveyor then visits all the grid intersect points by the most efficiently deemed route and records herbivore impact assessments at each point. This will require an initial assessment, once the location is reached, as to which habitat type the point should be assessed. There is no requirement to permanently mark plots as points can then be relocated using GPS.  Where previous HIAs have already been carried out on smaller estates with known georeferenced locations we recommend that these locations are resurveyed to ensure consistency and detecting changes in herbivore impact. 

At the sampling point, most indicators in the method should be assessed in a 2 m x 2 m quadrat area. Some indicators, such as tree seedling densities and/or herbivore impact on tree seedlings should be assessed in a wider 10 m radius area around the quadrat. 

For large estates and entire Deer Management Group (DMG) surveys, size from around 20,000 hectares (200 km-sq.) to 200,000 hectares (2000 km-sq.), we recommend the grid-cell and walk-through Rapid Assessment of Grazing and Trampling Impact method developed by the Macaulay Land Use Research Institute. Sampling selection is similar to the grid intersect method but involves surveying randomly selected ¼-km² squares (sample squares) stratified by management unit and vegetation type. Sample squares are based on the Ordnance Survey National Grid. The number of squares sampled for each vegetation category is a function of the total sample and the area of each vegetation category within each management unit. The total number of sample squares depends on the resource available, although minimum of 15% of the total area to be surveyed. 

The surveyor gets to the square and walks through recording indicators for dominant habitat types found within the square to assign an impact class of habitats walking through the square. In order to simplify the assessment: all impact classes which covered less than 25% of a square are ignored; the highest impact class is recorded if this covered more than 50% of the square; and an intermediate impact class is recorded if the highest class covered less than 50%, but more than 25%. Where vegetation mosaics are recorded as a unit, the impact class on blanket bog is given precedence over impacts on dwarf-shrub heath, which in turn take precedence over impacts on other habitat types. The overall impact class for a vegetation type is calculated by taking the mean of the impact classes for the sample squares in which the vegetation type occurred.

Timing of Survey

It is important to decide the timing of the HIA survey. Herbivore impact indicators are based on vegetative features that will vary seasonally and depend on the dominant habitat type and herbivore of interest. For example, shrub and tree seedling browsing by deer will be strongest during the winter, in contrast, smooth grassland deer and sheep browsing impacts may be highest at the late summer. The timing of the survey should reflect the estate managers herbivore(s) of interest and dominant habitat type. For many estates interested in detecting deer browsing impacts we suggest April and early May to detect browsing overwinter before the new seasons’ growth masks the previous seasons impact. 

Indicators

Below are Herbivore Impact Assessment indicators for four dominant open hill range land habitats. Each habitat indicator has a code (e.g., BB1) that corresponds to an evidence statement in the technical annex. In total, there are 26 indicators across habitat types: 5 for blanket bogs, 7 for dwarf shrub heath, 6 for tussock grassland and 8 for smooth grassland. 

Blanket bog

Disturbance

Browsing and grazing

Species composition

Other indicators to note for blanket bog:

(1) Prescence and browsing of dwarf birch (Betula nana); (2) Type of herbivore dung present (deer, sheep, cattle, hare); (3) Tree/shrub species and seedlings present.

Dwarf shrub heath              

Disturbance 

*Surveyors should note if areas have been subject to muirburn or wildfires

Browsing and grazing

*Surveyors should note if areas have been subject to muirburn or wildfires, as this can form carpet growth-forms

Species composition

Other indicators to note for dwarf shrub heath:

(1) browsing of bog myrtle (Myrica gale); (2) stem breakage of heather (Calluna vulgaris); (3) Type of herbivore dung present (deer, sheep, cattle, hare); (4) Muirburn/burning

Tussock grassland

Disturbance 

*Surveyors should note if areas have been subject to muirburn or wildfires

Browsing and grazing

Species composition

Other indicators to note for tussock grassland:

(1) Signs of grazing of less preferred species other than tussock formers, e.g., rush (Juncus spp.), thistle (Cirsium spp.), heath bedstraw (Galium saxatile), tormentil (Potentilla erecta) and mosses;  (2) Presence, flowering and grazing of tall herbs (e.g., Meadowsweet [Filipendula ulmaria], Devil's bit scabious, [Succisa pratensis], wild angelica [Angelica sylvestris]); (3) Hard fern (Blechnum spicant) browsing on leaves; (4) spatial extent of heather (continuous, patchy, sparse); (5) Type of herbivore dung present (deer, sheep, cattle, hare); (6) Muirburn/burning

Smooth grassland                                                                

Disturbance

Browsing and grazing

Species composition

Other indicators to note for smooth grassland:

(1) Type of herbivore dung present (deer, sheep, cattle, hare)