64
Appendix 4: Technical report Developing models to estimate the occurrence in the English countryside of Great Crested Newts, a protected species under the Habitats Directive [WC1108] End-user testing – to test model applications and fitness for purpose Summary The views, knowledge, understanding and attitudes of potential end- users towards predictive modelling were assessed using three approaches: (1) an online questionnaire; (2) a workshop for a wide range of stakeholders attending the Herpetofauna Workers’ Meeting; and (3) a workshop for professionals working in local and national government agencies, consultants, developers and utility providers. The vast majority of end-users were receptive to the idea of predictive modelling being used in decision-making for great crested newts, although there were some qualifiers and reservations. The main areas identified as having potential for applying SDMs were status assessment; risk assessment; impact assessment; assessing connectivity; informing planning and licensing decisions; targeting surveys; improving the outcomes and cost-effectiveness of surveys; habitat creation and management; and development mitigation. Whilst the general concept of predictive modelling is widely understood by end-users, there is a need to raise awareness of its value in particular applications and the attendant limitations. Concern was expressed over the reliability of model outputs and predictions, and whether the levels of confidence associated with predictions were appropriate for due-diligence and decision-making. Indeed, at a site-specific level modelling is no substitute for on-the-ground surveys. Implementing a modelling protocol would clearly require widespread training and capacity building, and the uptake may be easier for large organisations with the relevant infrastructure already in place. Considerable concerns were expressed over existing data management and data flow systems, which – in their current state of functionality – would not be able to efficiently provide the data that models require in terms of quality or quantity. Indeed, there was a risk that both input data and model outputs 1

sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

  • Upload
    lythien

  • View
    213

  • Download
    0

Embed Size (px)

Citation preview

Page 1: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Appendix 4: Technical report

Developing models to estimate the occurrence in the English countryside of Great Crested Newts, a protected species under the Habitats Directive [WC1108]

End-user testing – to test model applications and fitness for purpose

Summary

The views, knowledge, understanding and attitudes of potential end-users towards predictive modelling were assessed using three approaches: (1) an online questionnaire; (2) a workshop for a wide range of stakeholders attending the Herpetofauna Workers’ Meeting; and (3) a workshop for professionals working in local and national government agencies, consultants, developers and utility providers. The vast majority of end-users were receptive to the idea of predictive modelling being used in decision-making for great crested newts, although there were some qualifiers and reservations. The main areas identified as having potential for applying SDMs were status

assessment; risk assessment; impact assessment; assessing connectivity; informing planning and licensing decisions; targeting surveys; improving the outcomes and cost-effectiveness of surveys;

habitat creation and management; and development mitigation. Whilst the general concept of predictive modelling is widely understood by end-users, there is a need to raise awareness of its

value in particular applications and the attendant limitations. Concern was expressed over the reliability of model outputs and predictions, and whether the levels of confidence associated with predictions were appropriate for due-diligence and decision-making. Indeed, at a site-specific level modelling is no substitute for on-the-ground surveys. Implementing a modelling protocol would clearly require widespread training and capacity building, and the uptake may be easier for large organisations with the relevant infrastructure already in place. Considerable concerns were expressed over existing data management and data flow systems, which – in their current state of functionality – would not be able to efficiently provide the data that models require in terms of quality or quantity. Indeed, there was a risk that both input data and model outputs could be misused, both intentionally and unintentionally. There was no consensus over whether modelling might be best implemented through a local, regional or a centralised national system. Whatever options are proposed for modelling, there is a need for governance to be clear and authoritative, with common standards. Changes to legislation, policy and guidance may go some way to achieving this.

1

Page 2: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

4.1. Introduction

Incorporating predictive modelling into great crested newt conservation requires consultation with potential end-users and the wider conservation community. This would be needed with any significant amendment to policy or practice, but – depending on the precise applications and implementation mechanisms – it will be particularly important for modelling since it could involve a fundamental change to existing practice. Moreover, there might be a need for substantial shifts in technical competence, procedure, regulation, and the practice of key routine activities including newt survey and mitigation.

We investigated end-user and conservation community needs via three main approaches:

An online questionnaire of selected key potential end-users (January - February 2015); A large workshop at a conference for a range of practitioners and volunteers in the

amphibian conservation community (the Herpetofauna Workers’ Meeting, Newcastle, February 2015);

A targeted workshop for representatives from key end-user sectors (London, June 2015).

The online questionnaire sought views on broad issues, such as the degree to which those involved in great crested newt decision-making are aware of modelling, and understanding of its potential benefits. We also asked detailed questions about the specific needs of users. The online questionnaire also sought expressions of interest in further participation in the project, as well as requesting data and case studies. In the Herpetofauna Workers’ Meeting, we aimed to assess the views of a wider cross-section of potential users of modelling, as well as those who would not use modelling but are still involved in great crested newt conservation. Finally, the London workshop aimed to explore the views of a focused group of potential end-users from different professional sectors, based on interim results obtained up to the end of June. This allowed a more thorough assessment of some key issues, and also allowed us to specifically recruit construction industry representatives, since this sector was slightly under-represented in the two earlier consultation exercises.

In addition to these three specific exercises, we also gauged the responses and needs of potential end-users via less structured means. These included one-to-one meetings between the DICE modelling team and end-users in July/August 2015; ongoing liaison with individuals who demonstrated particular interest from the formal consultations; and discussions at the Natural England Great crested newt Pilot Advisory Group meetings.

When undertaking consultations we emphasised that the contract was to explore the issues, and not define a pre-determined policy shift.

4.2. Methods

4.2.1. Online questionnaire

The questionnaire was designed to address specific issues regarding the requirements of potential end users. An initial set of questions was drawn up and circulated within the project team, DICE staff involved with questionnaire design, and the project steering group for refinement. Following this a draft version of the questionnaire was tested using the online platform SurveyMonkey

2

Page 3: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

(http://surveymonkey.net). This was then piloted within the project team and then further refined before being distributed to the consultees.

The final version of the online questionnaire consisted of 11 pages (see Annex 1 for structure).

Most responses are shown as simple counts of multiple choice answers with percentages. We allowed an “other” option with free text where we considered it likely consultees might have responses in addition to our standard answers. For questions E2, E3 and F1 the consultees were asked to rate the importance of certain issues; for these questions we give an average rating, which is the average of the scores from 1 (not important) to 5 (very important).

The number of initial consultees for the questionnaire was limited to 24. Although the intention was to send the questionnaire directly to a larger group of consultees, the project steering group advised that this was not permissible within government survey protocols without significant additional investigation, which would have compromised the timing of the project. The 24 consultees were selected from a shortlist of potential respondents whose job, at least in part, involves making decisions related to great crested newts. The final list of consultees was selected in order to include as wide a range of sectors as possible. The 24 initial consultees were drawn from the following sectors:

Local Planning Authorities and associated contacts: 6 consultees Licensing authorities, statutory conservation advisors and other regulators: 6 consultees Construction industry: 2 consultees Utilities and major infrastructure providers: 2 consultees Ecological consultants: 3 consultees Data collectors and data providers: 2 consultees NGOs and civil society: 3 consultees

Each consultee was sent an email invitation to complete the questionnaire containing a brief introduction to the project, the background to the questionnaire and a link to the SurveyMonkey questionnaire. Consultees were encouraged to forward the link to colleagues and contacts. The consultation started took place from January to February 2015.

4.2.2. Herpetofauna Workers’ Meeting workshop

The objective of the workshop was to assess how a modelling approach to great crested newt conservation, might be received by a wider stakeholder group. We chose the Herpetofauna Workers’ Meeting as a forum for this assessment, as it has a diverse attendance, and because workshops are a traditional feature of the programme. Delegates at the meeting are representative of the typical stakeholders likely to be interested in modelling, including scientists, consultants, statutory agency staff, volunteers and NGO representatives. They therefore represent potential end-users as well as others with roles closely linked to great crested newt decision-making and data collection.

The workshop was advertised in advance as part of the promotion for the Herpetofauna Workers’ Meeting. Those registering for the meeting were given the choice of opting for the workshop, from a choice of four. The meeting took place on 7-8 February 2015 at the Vermont Hotel, Newcastle-upon-Tyne, and was organised by Amphibian and Reptile Conservation (hereafter, ARC) in conjunction with Amphibian and Reptile Groups UK (hereafter, ARG UK). The programme includes a variety of talks on amphibian and reptile conservation as well as a choice of participatory workshops.

3

Page 4: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Jim Foster (ARC) and Richard Griffiths (DICE) led on preparing the workshop format and content, with input from Dimitrios Bormpoudakis (DICE) and Thomas Starnes (ARC), and all four delivered the workshop. Liam Russell (ARC) and Jim Foster led on the post-workshop reporting.

Participants signing up for the Herpetofauna Workers’ Meeting were asked to select two of four workshops run during the event. The information provided to potential registrants included the following description of the great crested newt modelling workshop:

Workshop A: Towards a risk-based approach for great crested newt conservation?

Leaders: Richard Griffiths (DICE) & Jim Foster (ARC)

The locations of most great crested newt ponds is unknown, making for complex decision-making. DICE, ARC and the Freshwater Habitats Trust are collaborating on a Defra-funded project to scope a new approach to conserving newts. We’re developing models to predict where newts occur, and assessing risk based decision-making. Underlying this approach is a focus on conserving populations rather than individuals. In this workshop we’re seeking practitioner reaction to our ideas. How would this new approach impact on surveys for development, mitigation planning, or pond creation schemes? Can you suggest any improvements to our proposals? ARG members, local authority staff and consultants are especially encouraged to participate.

One hundred and seven out of a total of 169 registrants signed up to participate within the workshop. It was the most popular workshop of the four offered at the conference.

Two workshop sessions were held:

Session 1: Saturday 7 February 2015 (51 registered participants) Session 2: Sunday 8 February 2015 (56 registered participants)

Both sessions were led by Richard Griffiths (DICE), Jim Foster (ARC), Thomas Starnes (ARC) and Dimitrios Bormpoudakis (DICE). The sessions lasted 1 hour 30 minutes each.

The workshop started with a brief introduction to species distribution modelling, and an exploration of how this could be applied to great crested newts. We then briefly demonstrated two potential practical applications of modelling:

(1) Optimising survey effort when assessing the impacts to a hypothetical pipeline development, and

(2) A case study of modelling being used to inform land use allocation by a Local Planning Authority (based on an ARC project with Wrexham County Borough Council).

Following this introduction the participants were asked to form small groups (with approximately 7-12 members) to discuss the value of using predictive models for great crested newts in one of five scenarios (approximately 40 minutes, during which time the workshop leaders circulated to assist with queries arising from the discussions):

Scenario 1: Modelling as a tool in impacts assessment and mitigation planning Scenario 2: Modelling as a tool in survey planning Scenario 3: Modelling as a tool in Local Planning Authority development control

(determining planning applications)4

Page 5: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Scenario 4: Modelling as a tool in Local Planning Authority strategic planning (allocating future land use)

Scenario 5: Modelling as a tool in planning habitat creation & management

For each scenario, participants were asked to consider the following points and record their written comments in a table on a prepared reporting sheet:

Advantages: What are the potential benefits of modelling for this application? Disadvantages: What are the possible downsides or risks of modelling for this application? Barriers to implementation: What factors would make it more difficult for modelling to be

used in practice? Potential users: which organisation(s) do you think might use models for this application?

(please tick & add more if desired) NE/NRW/SNH Local Planning Authority Ecological Consultant Developer University Landowner Local Records Centre Volunteer surveyor ARG National NGO (e.g. ARC) Local NGO (e.g. Wildlife Trust) Other (please state):

General comments

After the small groups had discussed their responses to the questions, a spokesperson from each group presented the outcome of their discussion to the whole workshop (approximately 20 minutes). Given timing constraints, the workshop leaders did not respond in detail to individual comments made during the feedback sessions. The reporting sheets were collected for later transcription.

Towards the end of the workshop, we gave a brief summary of the next stages for the project, asked participants if they had any datasets we might use in model development, thanked them for their time, and noted that a project report would be due in autumn. We stressed that the modelling and risk-based approaches under discussion were still being explored, and that government had not yet decided that this would be pursued. We explained that the workshops they had helped with were part of that exploration.

We ended the workshop with a “straw poll” by asking participants the following question: “Would you eventually like to see predictive modelling included as a fundamental aspect of great crested newt conservation policy, planning and guidance?”, with possible responses: Yes, No, Don’t know.

4.2.3. Targeted end-user workshop

Invitations to this workshop were based on liaison with the Steering Group, the Natural England-led Great crested newt Pilot Advisory Group, feedback from the online questionnaire, and discussions

5

Page 6: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

held at the Herpetofauna Workers’ Meeting workshop. We invited 59 representatives from the following sectors:

Developers/Utilities companies Ecological consultants Government agencies Local planning authorities (LPAs) NGOs and recording groups.

The workshop arrangements were agreed with the Steering Group in advance, and invitations agreed by email with a maximum of 25. The workshop was held at Nobel House, Defra, in central London, from 11.00 to 16.00 on 30 June 2015.

The workshop structure was partly based on our successful experience with the workshop at the Herpetofauna Workers’ Meeting (see above). The first section set the scene and provided background to modelling and preliminary results obtained so far. This workshop, however, was designed to elicit slightly different information given that we were able to select a more targeted audience, and we had interim results from the modelling exercises. We discussed the interim results of the modelling work done using environmental DNA (eDNA) surveys in Kent, Cheshire and Lincolnshire. We demonstrated the different model types, discussed model performance in different regions and at different scales, and briefly explained the data, hardware, software and expertise requirements. This was followed by a general discussion.

We then divided the participants into six break-out groups by sectors. One member of the contract team facilitated each break-out group for a 45 minute period. During this facilitated discussion, we asked each group to discuss and report on the following using a structured proforma:

Applications (Please list possible applications of a modelling approach within your sector) Advantages (What are the benefits of modelling for this application?) Disadvantages (What are the possible downsides or risks of modelling for this application?) Implementation (What might be needed to implement modelling for this application (e.g.

training, funding, hardware etc)?) General comments.

One representative from each group provided oral feedback at the end of the session. Groups were also asked to note any related issues that arose during the course of their discussions on post-it notes, and these were collated at the end of the workshop. Each sector was also asked to provide a response to the question: “Does your group think that a modelling approach could help decision-making within your sector?” with three possible responses: Yes, No, Don’t know.

Following this we held a discussion on workshop conclusions, and finally a brief section in which we explained the next steps in the project, with an offer for one-to-one modelling meetings and a request for case studies.

4.3. Results

4.3.1. Online questionnaire

A total of 41 responses was received, four of which were mostly blank and therefore removed from the analysis. The remaining 37 responses were included in the analysis.

6

Page 7: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Questionnaire Section A: User details

Question 1: Name

This was provided by 34/37 respondents (92%)

Question 2: Organisation

This was provided by 33/37 respondents (89%)

Question 3: Position

This was provided by 33/37 respondents (89%)

Question 4: E-mail

This was provided by 34/37 respondents (92%)

Question 5: Contact telephone number

This was provided by 31/37 respondents (84%)

Question 6: What sector do you work in?

This was answered by 37/37 respondents (100%)

Table 1. The different sectors represented at the workshop.

Sector Response count Percentage

Local Authority 12 32.4Statutory Agency 15 40.5Development (including construction, infrastructure, utilities, etc) 4 10.8Ecological/Environmental Consultancy 3 8.1NGO 3 8.1Other (please specify) 0 0

Question 7: How much of your job is concerned with decisions relating to great crested newts?

This was answered by 37/37 respondents (100%)

Table 2. Time spent dealing with decisions relating to great crested newts.

Sector Response count Percentage

Not at all 2 5.4About 25% 25 67.6About 50% 4 10.8About 75% 5 13.5All of my job 1 2.7

Questionnaire section B: Why do you need data on great crested newts?

Question 1: Why do you need information on great crested newts to fulfil your job responsibilities? Please tick all that apply

7

Page 8: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

This was answered by 37/37 respondents (100%)

Table 3. Utility of information on great crested newts to fulfil job requirements.

Answer options Count Percentage

To inform land use planning, such as development control or forward planning 22 59.5To assess whether a plan or project would negatively affect great crested newts 28 75.7To assess how a conservation project could benefit great crested newts 17 45.9To assess whether a survey for great crested newts is needed, or to specify such a survey 24 64.9

To design a mitigation project for great crested newts 17 45.9I don't 2 5.4Other (please specify) 8 21.6

“Other” responses:

Information helps me to keep project teams up to date and understand why mitigation measures are required on some of our projects.

I interpret findings of GCN survey information for purpose of determine impacts requiring licensing.

To determine whether a licence can be granted - and whether the 3 licensing tests can be met. To advise on licensing requirements in relation to GCN. To advise other user needs

I am basing the answers to these questions on my experience in my previous post in NE as wildlife adviser.

To assess whether mitigation measures submitted in support of a planning application are appropriate.

To inform the design of a possible local biodiversity offsetting approach towards GCN To inform Government-funded strategic research into GCN status and monitoring,

aka Evidence Enhancement Project (EEP). We were a project partner in Phase 1 with Hyder-Cresswell, from early stages of development onwards, and supplied many field surveyors. We were partner in Phase 2 as well, and supplied/managed most of the field surveyors.

To assess status and conservation needs.

Question 2: What are your main data need(s) related to great crested newts? Please tick all that apply

This was answered by 37/37 respondents (100%)

Table 4. Main data needs of respondents on great crested newts.

Answer options Count Percentage

Distribution 27 73.0Habitat quality information 28 75.7Population size 28 75.7Probability of presence 31 83.8

8

Page 9: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Information about impacts 28 75.7Trends 11 29.7None 2 5.4Other (please specify) 3 8.1

“Other” responses:

All of the above information is required as part of a mitigation licence application Use of mitigation or compensation measures created for GCN. Post development

monitoring and effectiveness of the compensation plus what remedial actions will help resolve any issues identified with these habitats.

Occupancy rates, pond densities, HSI data. We need new qualitative assessment techniques for terrestrial data. I have attempted such a thing in Phase 1 of EEP, where presence of important habitat types nearby (woodland, scrub, rough grass etc) are scored out of 10. We collected this data for all ponds in Phase 1, but I don't think [name redacted] used it in the report! I don't have access to the data, so can't publish it myself, which is a shame.

Questionnaire section C: How do you currently meet these data needs?

Question 1: In your job, what are your main sources of information about the status and distribution of great crested newts? Please tick all that apply

This was answered by 37/37 respondents (100%)

Table 5. Sources of data on great crested newts.

Answer options Count Percentage

Online resources (e.g. NBN Gateway) 13 35.1Published guidance notes 12 32.4Books/scientific literature 11 29.7Third party (e.g. NGOs, local Amphibian and Reptile Groups) 20 54.1Local Record Centres 24 64.9Bespoke surveys 27 73.0Professional advice 17 45.9None 0 0.0Other (please specify) 5 13.5

“Other” responses:

Client specifies ecological constraints on projects, which may include GCN measures It is the applicant for a GCN mitigation licence who is responsible for gathering and

evaluating any information on the species, it is my job to assess if it has been done accurately and in accordance with published guidance.

Licence application details Survey/mitigation schemes submitted in support of planning applications

9

Page 10: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

I interpret the landscape from maps, aerial photographs, and underlying geological knowledge. Clayey areas are best for GCN ponds.

Question 2: At the moment, when you need information on great crested newts, how quickly do you typically need to obtain it? Please select one option

This was answered by 37/37 respondents (100%)

Table 6. How quickly respondents require information on great crested newts.

Answer options Count Percentage

Within 1 week 7 18.9Within 1 month 12 32.4Within 6 months 5 13.5Within a year 0 0.0Not applicable 6 16.2Other (please specify) 7 18.9

“Other” responses:

Prior to determination of planning application This is usually seasonal information and will depend on the time of year. If the

likelihood of GCN is reasonable, insufficient survey information in the planning system may result in the need for surveys which cannot be undertaken for up to 7 months, in order to ensure compliance with the Habitats Regulations, NERC duty and Prevention of Crime act.

Usually statutory 21 days for consultation response to LPA Depends - variable It depends on project timescales but we would typically be looking at months not

weeks. LRC provide annual updated buffered protected species GIS layer that includes info

on GCN distribution. Less than a week quite often. Clients don't like waiting a couple of weeks for LRC

searches.

Question 3: How strongly do you agree with the following statements? Please tick one box per row

This was answered by 37/37 respondents (100%).

10

Page 11: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Strongly agree

Agree Neither agree nor disagree

Disagree Strongly disagree

N/A0

2

4

6

8

10

12

The seasonal nature of great crested newt survey data collection is not a problem for my work

The cost of great crested newt sur-vey data collection is a not problem for my work

Figure 1. How strongly respondents agree with the following statements; (blue), the seasonal nature of great crested newt survey data collection is not a problem for my work, and (red) the cost of great crested newt survey data is not a problem for my work.

Questionnaire section D: Awareness of predictive models

Question 1: Have you ever used predictive modelling to look at species distribution or status before?

This was answered by 36/37 respondents (97%)

Yes: 10 (27.8%) No: 26 (72.2%)

Question 2: If yes, which model have you used?

This was answered by 10/37 respondents (27%):

ARCH Model MaxEnt models through ARC contract. My own own predictive model for Harbour Porpoise in ArcGIS for MSc dissertation ENMs, HSMs, GLMs, AUCs - too many to list Unsure what it is called. Individual based population models linked to a GIS to model species distribution over

space and time Habitat Suitability Index MaxEnt Own Habitat Suitability

Question 3: Were there any benefits to using the model over traditional methods?

This was answered by 14/37 respondents (38%):

Yes: 10 (71.4%) No: 4 (28.6%)

Please describe any benefits to using the models:

11

No.

of r

espo

nses

Page 12: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Useful broad background information Gave a better estimate of national distribution and numbers They are very accurate Helpful in predicting occurrence and looking for trends in relation to temporal

patterns & habitat availability. Each model has its own benefits and limitations - it is up to the researcher to

determine whether the limitations are relevant / acceptable given the system under study.

Can provide decision maker with information on the probability of finding GCNs at a particular locality. Can also be used to assess the suitability of localities for mitigation, offsets etc.

Practical difficulties and cost of obtaining field data. Also looking at predicting disease spread.

It provided information about the likely terrestrial distribution which is very difficult to measure through surveys

Question 4: Were there any problems to using the model?

This was answered by 13/37 respondents (35%):

Yes: 7 (53.8%) No: 6 (46.2%)

Please describe any problems you had using the models:

Does not prevent the need for detailed site assessment and survey Wide confidence intervals on national estimates Always going to be some inherent error in modelling and output may require expert

interpretation. Good baseline data. Knowing what to set variables at, e.g. maximum movement of

an animal, fecundity etc As with any model it is based on a series of assumptions that individual animals

ignore. Developing expertise required. Cost and time in gathering and processing

environmental and newt data for the model.

Question 5: How strongly do you agree with the following statement: ‘I have an understanding of predictive modelling applications for biodiversity conservation and what they can do’

This was answered by 35/37 respondents (95%)

12

Page 13: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Strongly agree Agree Neither agree nor disagree

Disagree Strongly disagree0

2

4

6

8

10

12

14

16

18

Figure 2. How strongly respondents agree with ‘I have an understanding of predictive modelling applications for biodiversity conservation and what they can do’.

Question 6: On the basis of what you what you know now, how would you rate the potential value of using predictive modelling for your decision-making? Please select one option

This was answered by 36/37 respondents (97%)

Not valuable at all

Little value Some value Very valuable Extremely valuable

0

2

4

6

8

10

12

14

16

18

Figure 3. The potential value of using predictive modelling for decision-making.

Questionnaire section E: Using predictive models

Question 1: If you were given a map that predicted the presence and absence of great crested newts, which of these levels of resolution would help you in your job? Please tick any that apply

This was answered by 36/37 respondents (97%)

13

No.

of r

espo

nses

No.

of r

espo

nses

Page 14: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Table 7. Required map resolution for predicting presence-absence of great crested newts.

Answer options Count Percentage

10 km grid square 3 8.3between 10 km and 1 km 4 11.11 km grid square 15 41.7Better resolution than 1km grid square (e.g. 100 m) 30 83.3Individual pond 27 75.0

14

Page 15: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Question 2: If you could actually use a predictive model in your work, how important would each of the following features be? Please select one option per row

This was answered by 36/37 respondents (97%)

Table 8. Importance of features of predictive models.

Answer Options Not important at all

Little importance

Some importance

Very important

Extremely important

Rating Average

Ease of use 1 1 6 12 16 4.14Ease of understanding model outputs

1 0 2 13 20 4.42

Ease of communicating model outputs

1 0 5 14 16 4.22

Demonstrable high reliability of model predictions

1 0 4 10 21 4.39

Identification of the relative importance of particular factors in predicting great crested newt presence

1 2 9 11 13 3.92

Access to the model itself (rather than just outputs from the model)

2 8 13 10 3 3.11

Access to a map of predicted presence/absence derived from the model (static map, e.g. a jpeg or PDF)

1 0 9 14 12 4.00

Access to a map in the form of a GIS layer of predicted presence/absence, derived from the model

1 5 5 10 15 3.92

Access to an online system that provided information on newt presence and impact after you enter the grid reference for the area concerned

1 2 6 13 14 4.03

Access to a person (via phone, email or in person) who would run and interpret the model for you

4 11 12 8 1 2.75

Cost benefit analysis to support use of a model, compared with traditional surveys

8 8 9 6 5 2.78

“Other” responses:

I am unlikely to be using a predictive model myself, but outputs from the model would direct mitigation measures on construction projects

With regards to reliability of model predictions, the key thing to know is the level of accuracy of predictions and associated uncertainty (confidence intervals) so we know how much the outputs can be relied on.

don't see model replacing survey work. Very naive to think it would. Useful to support survey work.

15

Page 16: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

It would be expected that the applicant would run the model and submit as part of the application - NE would assess the outputs

It would be important to know that the model would be reviewed at regular intervals to ensure that the model is based on up to date data.

Ability of model to predict impacts of development on local FCS Heat mapping for population sizes, pond densities, occupancy rates etc.

Question 3: How important are the following outputs that might be obtained from a predictive model in your work? Please select one option per row

This was answered by 36/37 respondents (97%)

Table 9. Importance of outputs obtained from predictive models.

Answer Options Not important at all

Little importance

Some importance

Very important

Extremely important

Rating Average

Reliability of determining presence of great crested newts at a particular location

0 0 3 9 24 4.58

Reliability of determining absence of great crested newts at a particular location

1 0 3 8 24 4.50

The abundance of great crested newts at a particular location (population size)

1 3 8 14 10 3.81

The importance of a particular parcel of land (whether or not it contains ponds) for great crested newts

0 1 7 10 18 4.25

The importance of a landscape for great crested newts

0 2 8 16 9 3.91

Likelihood of a particular project impacting great crested newts

0 0 4 8 24 4.56

Assessment of the scale of an impact of a particular project affecting great crested newts

0 1 3 11 21 4.44

Location of suitable areas for mitigation works to compensate for development

1 1 5 11 18 4.22

Presence of barriers to dispersal to great crested newts in a particular area

0 3 5 13 15 4.11

Conservation targets for great crested newt at a local or regional scale

1 6 11 10 7 3.46

Conservation targets for great crested newts at a national scale

1 10 14 8 3 3.06

“Other” responses:

16

Page 17: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Ability of the model to ensure compliance with Habitats Regulations, NERC act and Prevention of Crime act

Again, it is very important that uncertainties associated with any modelled output is understood.

Questionnaire section F: Operational issues that may constrain embedding models in decision making

Question 1: How important are the following issues in terms of how you would implement predictive models in your work on great crested newts?

This was answered by 36/37 respondents (97%)

Table 10. Importance of different issues in implementing predictive models when working with great crested newts.

Answer Options Not important at all

Little importance

Some importance

Very important

Extremely important

N/A Don’t know

Rating Average

Staff time 0 1 6 12 13 4 4.16Staff expertise 0 1 6 12 14 3 4.18IT facilities 0 1 9 9 11 6 4.00The scale of benefit of using a model compared to existing methods

0 1 4 17 12 2 4.18

Ensuring models and their outputs are acceptable to other bodies/organisations

1 0 4 10 20 0 4.37

Ensuring there are no concerns relating to data ownership and responsibilities

1 2 9 8 14 2 3.94

Ensuring model outputs and uncertainties are effectively communicated

0 0 4 11 19 2 4.44

Ensuring model outputs comply with policy and procedural restrictions related to land use planning

1 2 5 9 18 1 4.17

Ensuring model outputs comply with policy and procedural restrictions related to protected species legislation

1 2 2 5 24 2 4.44

“Other” responses

17

Page 18: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

My focus would be on the outputs from models rather than use of the models

Questionnaire section G: Request for data and case studies to assist with model development

Question 1: May we contact you about datasets or case studies that might help us?

This was answered by 35/37 respondents (95%)

Yes, I have dataset I may be willing to provide: 1 (2.9%)

Yes, I have case studies that may be useful: 10 (28.6%)

No: 24 (68.6%)

Questionnaire section H: Comments and further involvement

Question 1: Please provide any other comments or views you may have on the potential application of models

This was answered by 7/37 respondents (19%):

There would be a need to ensure Elected Members and the public in general that models are an appropriate alternative. Also would their used be Judge/JR proof?

As a construction contractor, our surveys are done by third-party ecologists, I will pass this on to them and hope they will provide you with relevant data sets.

Assessing impacts from on-going activities, e.g. Highway runoff contaminated with salt, to determine whether they are affecting population levels.

In my experience in 2 years of local authority planning I cannot think of a single situation whereby a local planning authority has expressed concerns over the impact of GCN survey/mitigation requirements on development (or the growth agenda as it is referred to here). In my experience in LPA planning and ten years in the commercial sector, concern is generally raised by developers in respect to the perception that GCN survey/mitigation places an unnecessary burden on planning. This is not my experience provided that adequate preparation has gone into preparing planning applications. This attitude is common to all legally-protected species e.g. bats. I can cite examples where GCN have been recorded within supposedly suboptimal habitat and therefore a concern I would have over the use of predictive modelling (with its obvious limitations) is that it will inevitably discount sites that are in fact locally important in favour of obvious sites. It may lead to an over reliance on worst-case mitigation scenarios. The Conservation Regulations make it clear that impacts to individual GCN are an offence so the shift of focus onto solely national/regional population-level impacts may cause issues in terms of local conservation objectives.

My area is very well covered by GCN survey work and I am less interested in the predicting presence/absence and more interested in assessing the impact of developments and land use plans on local FCS of GCN

I am sceptical of the current drive for eDNA instead of traditional survey to save costs. I am fully in support of predictive modelling however, but why is the issue of cost-saving so upfront here? The mandatory mitigation that the SNCOs require is inordinately more expensive than presence-absence surveys, and we all know how

18

Page 19: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

limited the benefits often are. The offences are badly-designed, and until they're overhauled, the expense on fencing and inflexible excessive capture effort is insupportable. By the way, the case study I have is a minerals site in [redacted], where pond surveys showed zero GCN presence, but subsequent reptile survey and resultant terrestrial GCN survey showed major presence, at great and disruptive cost (£140k) to the developer.

Use in forward planning is important to avoid impacts in first place.

Question 2: Would you be interested in attending an end-user workshop (though please note that places will be limited so we cannot guarantee you a place at this stage)?

This was answered by 35/37 respondents (95%)

Yes: 27 (77.1%) No: 8 (22.9%)

4.3.2. Herpetofauna Workers’ Meeting workshop

The responses have been organised into general themes within each scenario. Every comment received during the workshop has been included below (with the exception of any that were illegible). After each quote, a number in brackets identifies which of the two sessions the response is from. Where there was more than one group covering the same scenario the number is followed by a letter to identify the group.

Scenario 1: Modelling as a tool in impact assessment and mitigation planning

Advantages:

The potential for the model to be used in a variety of stages in the impact assessment process was identified.

“Use for forward planning” (1) “Can be used to determine current conservation status/favourable conservation

status” (1) “Help plan survey, including targeting” (1) “Cumulative impact assessment” (2)

Both groups identified the potential for the model to save time and cost.

“Quick assessment” (1) “Save time and cost” (2)

One group thought that the objectivity of the model was an advantage. We assume that this is in comparison to traditional impact assessment methods, which are subjective to a certain extent.

“Objective assessment (more)” (1)

Both groups also thought the model would be useful in terms of identifying the importance of particular habitat features (such as an individual pond or block of terrestrial habitat) at a landscape level and therefore how potential impacts to these features could be addressed:

“Holistic approach across area” (2)

19

Page 20: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Identifying important ponds for connectivity” (1) “Retention of connectivity – tunnels etc” (2) “Avoidance of sensitive areas” (2)

Other miscellaneous advantages included possible other applications and the potential for the model to be updated:

“Live and can be updated” (1) “Better alternative to deregulation” (2) “Transferrable to other sectors use modelling” (2)

Disadvantages:

Both groups identified potential problems relating to the predictive power or precision of models. There was a particular concern about whether a model would work effectively in different geographical areas and whether it could lead to some populations being missed:

“Model not suitable in some areas” (1) “Different areas need different models” (1) “Overlooking ponds which currently isolated but which have previously been

connected and therefore important” (1) “Not real data – risk of missed populations” (1) “Needs adaptation, based on lowland – is it transferrable to upland?” (2) “Loss of range?” (2)

There were concerns about the potential for the model to be misused (either intentionally or unintentionally) and users not understanding the limitations of the model:

“Potentially seen as an alternative for survey” (2) “It is a risk based approach” (2) “Misused/not implemented in the correct way” (2) “Only focusses on GCN [great crested newts]. What if area is very important for

dormice?” (2)

There were also concerns about the resources required to create the model and how it could be accessed:

“Time cost and access to models” (1) “Licensing of software MAXENT use” (1)

Barriers to implementation:

There was limited overlap between the groups on what barriers they identified. One group was particularly concerned with the reliability of the data used to create the model and both had concerns about the transferability of the model between areas:

“Needs good survey as a baseline” (2) “Ground testing” (2) “Need to take account of variation in habitat requirements in different areas” (1) “Dependent on data, may not transfer to other areas. Kent Essex X” (2)

20

Page 21: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Both groups identified a variety of practicalities relating to how end-users would interact with the model, particularly in the level of knowledge required to accept and interpret the output:

“How does it work? (2) “Access” (1) “Accessibility?” (2) “Licensing” (1) “Standard for interpretation of results” (2) “How is it used?” (2) “Who pays? – User based?” (2) “What comes out of the model?” (2) “Who hosts it?” (2) “Many groups involved!” (1) “Easy access to all” (2) “TRAINING NEEDS” (2) “People are used to the HSI [Habitat Suitability index] and don’t like change” (1)

Miscellaneous comments related to the practical application of the model in relation to the strict legal protection of individual great crested newts (as opposed to populations):

“Legislation protects individuals” (2)

General comments:

General comments related to the need to involve stakeholders in creation of the model and ongoing accessibility issues, as well as the need for ongoing maintenance/improvement. Both groups also identified some areas where they thought the model would be particularly useful:

“Need to include all stakeholders (inc ARGs) from beginning” (1) “If there is a single approach to modelling across the country the R+D agency could

end up with a monopoly. Terms & conditions of access and intellectual property rights must be specified at the beginning of the process” (1)

“Needs regular updates once live: new roads/new barriers, new ponds/habitats” (2) “Biggest strength may be in deciding where to put new ponds for biggest +ve impact”

(1) “In theory, modelling can be used to inform the planning and strategic location of

development and therefore newt areas, however, piecemeal approach to development often means newts are translocated several times. In this situation the creation of ‘newt areas’ – and a focus for strategic management/creation + protection of areas could (and probably will) be ignored. This modelling therefore is not only an advantage, but a serious disadvantage when the process is not followed and developers get away with too much.

“Precursor to biodiversity offsetting? Good or bad?” (2)

Scenario 2: Modelling as a tool in survey planning

Advantages:

21

Page 22: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Both groups identified the potential for refining survey effort at different spatial scales, including from individual pond level to a broader landscape level:

“Helps focus survey effort” (1) “Help target survey effort, especially in a large area” (2) “Target areas in terms of assessing presence/absence where no known GCN ponds”

(2)

One group identified potential for use in targeting habitat improvement:

“Could identify areas suitable for habitat improvement” (1)

Disadvantages:

Both groups identified several points relating to situations where individual ponds/populations could be missed by the model, and the implications arising. This was of particular concern around potential instances where populations could occur in unusual situations that the model would not account for. These concerns over predictive power led to questions about whether the model would be worthwhile if it was still necessary to undertake traditional field surveys:

“Potentially missing isolated populations in unusual areas” (1) “Biasing results by only looking at good habitats” (1) “Does not replace visiting ponds – is producing a model a waste of time?” (1) “In terms of development, just using the model wouldn’t allow us to confidently rule

out ponds to survey” (2) “Requirement of ground-truthing for data, e.g. missing ponds” (2)

Additional concerns were identified relating to the quality of the data on which the model is based, and the ability of the model to cope with biological or environmental changes, which could affect great crested newt distribution:

“Dependent on quality of data in your area” (2) “Account of factors that cannot be measured. e.g. climate change, chytrid” (2)

The issue of model accessibility was also raised:

“The knowledge of the user” (2)

Barriers to implementation:

Cost was considered the primary barrier to implementation, both to produce the model in the first place and any costs for end-users. The lack of other resources (particularly input data) was also identified as a potential barrier:

“Amount of data needed to produce model” (1) “Cost of producing national model” (1) “Cost of obtaining model” (1) “Availability of resources and cost” (2)

22

Page 23: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

One group also questioned whether statutory bodies would accept the model. The context for this comment was not clarified, but we assume this is in terms of how statutory bodies would view decisions made on the basis of the model as opposed to field survey:

“Acceptance by Natural England?” (2)

General comments:

Positive comments were received as the potential of the model to be another tool in the newt surveyors’ toolbox, although this also implied that it would not replace traditional survey methods. Ideas for how the model could be made available were also made:

“Used in conjunction with other survey methods (i.e. HSI, presence/absence) would be a useful tool” (2)

“If available on a site like MAGIC [Multi Agency Geographical Information for the Countryside, a government website] where professionals in ecology are quite familiar” (2)

Scenario 3: Modelling as a tool in Local Planning Authority development control (determining planning applications)

Advantages:

Both groups identified several areas where modelling could be used for efficiency savings in terms of identifying the likelihood of great crested newts being present within an area enabling better targeting survey effort and therefore reducing costs and time:

“Allows initial sift of risk” (1) “Enables survey to be targeted” (1) “Having a model can be used to focus survey effort” (2) “Reduces survey effort – ‘hot spots’” (2) “Add predictors to use in future p.a.’s [planning applications]” (2)

Both groups could particularly see advantages in using modelling to inform habitat creation measures, particularly in reference to mitigation projects, but also as part of conservation projects:

“Save money – alleviate need for public enquiry [sic] mitigating impact at first level” (1)

“LA’s [Local Authorities] can use it as a tool” (1) “Allows for joined up habitat/mitigation measures” (2) “Better informs mitigation” (2) “Enables a look at connectivity” (2)

Both groups could see advantages for Local Authorities in fulfilling their duties with regard to ensuring great crested newts are not adversely affected by proposed development. This was considered from the point of view of assessing impact as well as justifying why great crested newts need to be considered through the application process:

“Enables a look at connectivity” (2)

23

Page 24: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Having a model makes it easier to make a case for conservation. Speak authoritatively.” (1)

“LA’s can use it as a tool” (1) “Useful visual representation for non-experts” (2)

One group identified the cross-species potential of modelling:

“Can be used for different species e.g. bats/herps” (1)

Disadvantages:

Both groups were concerned about the amount of data required for the model, whether it existed in sufficient quality to allow accurate predictions. In addition there were concerns over whether the same model would be applicable in geographic different areas. However, one group also identified the potential for different Local Authorities to use different model as a disadvantage:

“Need the data set” (1) “Data management – local schemes v. national levels – fragmentation” (1) “’Rubbish in rubbish out’ model needs validation” (1) “Needs good baseline data locally” (2) “Different areas need different models” (1) “Model not necessarily applicable a local level” (2) “Must encourage LA’s to use a standard model” (1)

Both groups were also concerned over the potential for the model to be used as a substitute for more traditional surveys and that this would result in some ponds/populations being undetected:

“Would want to survey every pond as model not realistic” (1) “Potential misuse of model. Reading it as gospel” (2) “Using it as a replacement for real life survey” (2) “Tendency to over-rely on model” (2) “Chance of false dismissal of GCN ponds” (2)

One group was concerned about the use of the software by end users:

“Need [the data set] and to use the software (can it be done centrally?)” (1)

Barriers to implementation:

Both groups identified issues with the data required to create the model. There were concerns over availability of data in the first place, as well as confidentiality issues that could affect how the models could be used and their outputs disseminated:

“Lack of initial data” (2) “Lack of local data” (2) “Continue updating must happen – valid dataset” (1) “Data protection v. data accessibility” (1) “Consultancies hold onto data – NE [Natural England] – public domain” (1)

Both groups also identified issues relating to the skills required by end-users to use the model and interpret its output:

24

Page 25: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“ARGs [Amphibian & Reptile Groups] don’t have expertise, e.g. using software. Need a central body” (1)

“LA’s without an ecologist” (1) “Lack of local expertise to interpret and critique” (2)

General comments:

Both groups used the general comments section to expand on topics covered in the disadvantages and barriers sections. Issues around data confidentiality, the application of the model at a local level and the expertise needed to use the model were covered. One other comment about the technical aspects of creating the model was also received:

“’Secret’ vs public domain as we’re paying for directly/indirectly. Can surveys be published?” (1)

“Can we locally tailor it?” (2) “Needs clear guidance as to how to use it” (2) “Weighing HSI parameters pond size/density most imp. – model layers” (1)

Scenario 4: Modelling as a tool in Local Planning Authority strategic planning (allocating future land use)

Advantages:

All of the groups identified the potential usefulness of the model in strategic planning for highlighting sites where newts may be a constraint as well as sites where they would not. In addition, one group considered the model may be useful in planning conservation measures at a landscape scale:

“Highlight potential ‘problem’ sites – gcn present – mitigation needed” (1a) “Highlight ‘non-problem sites’ no gcn – no constraints” (1a) “Visually – easier to view areas to avoid impacting gcn” (1b) “Useful for strategic planning (assessing impacts)” (2) “Predicting best use of resources to improve connectivity – enhancements/

compensation” (2)

The potential for modelling to be used further along in the planning process was also identified at the project design stage and impact assessment stage, and mitigation design:

“Used to liaise with developers at design stage” (1b) “Understanding larger developments” (2) “Fulfilment of Habitats Directive consideration of protected species” (2)

One group also noted potential cost savings of using a model as opposed to traditional surveys and the potential for the same techniques to be applied to other species as part of the same model:

“Cheaper than sending out ecologists to assess areas” (1b) “Link model with reptiles” (1b)

Disadvantages:

25

Page 26: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

There was variation between the groups in the disadvantages identified. Two of the groups were concerned with the data underlying the model:

“Data sources, data updates” (1a) “Lack of records to improve model for localities” (2)

There was also concern about a variety of potential limitations to the model including the overall accuracy of the model (particularly the potential for small populations being missed), the ability of the model to cope with unusual habitats and the inability of the model to consider multiple species:

“Model may have error – wasted time/cost” (1b) “Small scale/isolated populations could be ruled out/disregarded? (1b) “Presence/absence might not represent conservation significance” (2) “SDM [Species Distribution Model] may not be able to cope with high anthropogenic

disturbance areas” (2) “Only looking at GCN – not account for other species” (1b)

There was also concern about the potential for the model output to be misinterpreted including that the limitations of the model would not be understood by some (particularly those with limited knowledge of great crested newts). In addition, one group were concerned that the competing demands on planning authorities would lead to the model not being considered:

“Planners use this as ‘Gospel’” (1b) “Used as an alternative to real life surveying (no corroboration) - interpretation” (2) “Protected species SDM will only be one factor for the LPA [Local Planning Authority]

to consider” (2)

Two groups were concerned about the potential costs of creating and using a model, including the requirement for training:

“Costs – system, training, ongoing support” (1a) “Central system. Who funds it? Who maintains it?” (1a) “Cost in training of model – council HAVE to employ ecologist – maybe too costly for

small scale schemes” (1b)

There was also concern over whether the model would be accepted as a valid technique when subjected to legal challenges in a planning context and therefore would Local Authorities be too risk averse to use it:

“Accepted by county agencies” (1a) “Accepted by legal teams – ‘can we be certain…’ risk averse” (1)

Barriers to implementation:

All of groups were concerned about the amount of data require to create the model, although this was more in reference to habitat data rather than newt distribution data:

“Underlying data sources, newts/botany/hedges/ponds etc ‘ground truths’” (1a) “Need land access” (1b) “Need up to date data (habitats)” (1b)

26

Page 27: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Need to factor in former mitigation sites” (2)

There were also concerns regarding the resolution of the model, both in terms of geographical variation and the model’s ability of cope with the newts’ biphasic (terrestrial and aquatic) lifecycle:

“Would model be the same across LPA’s? – regional variance” (1b) “Need to consider entire lifecycle for species” (2)

Understanding the limitation of the model including whether the model’s predictions would stand up to legal scrutiny through the planning process was also a concern:

“Quantify the uncertainty, understand the model restrictions and limitations” (1a) “Creating an SDM that will hold up at appeal” (2)

The need for training of those running and interpreting the model was also identified, as was the issue of who would be permitted to use/access the model:

“Ease of use – off the shelf package, training of use?” (1a) “Training of interpretation of results – do users understand outputs, non-ecologists”

(1a) “Access to system for ‘3rd parties’ e.g. developers/consultants – cost of use” (1a)

Financial considerations where also identified as a potential barrier, both the initial creation of the model and any costs to end users:

“Access to system for ‘3rd parties’ e.g. developers/consultants – cost of use” (1a) “Funding at an LPA level” (2)

General comments:

General comments were received primarily relating to the practical elements of the use of the model, particularly funding, who would be responsible for carrying out modelling, and how it would be applicable at various spatial levels. One comment was received regarding the need for the involvement of various stakeholders:

“Who funds costs of model?” (1b) “As big a dataset as possible is required therefore co-ordination of dispersed records

needs funding” (2) “Who has responsibility of model?” (1b) “Can this model be scaled down” (1a) “What level is system. County/National” (1a) “Best practice stakeholder involvement and participation – Wales good, other

places?” (1a)

Scenario 5: Modelling as a tool in planning habitat creation & management

Advantages

Each of the groups identified how the model could be useful in efficiently identifying features of value for great crested newts and targeting areas for habitat creation that provide the most gain for the invested resources:

27

Page 28: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Identify hotspots within a large scale landscape” (1a) “Identify hotspots and target areas for survey” (2) “Focus/target funding to gain the best outcome for GCN” (1a) “Help use resources efficiently (financial)” (1a) “Saves money and resources” (2) “Good to make decisions of connectivity” (1a) “Connectivity predictions so we can focus on areas of good connectivity” (1b) “Prioritise resources” (1b) “Targeted effort, tailored to specific areas, more efficient” (1b) “Cost benefit analysis may be easier” (2)

A number of potential applications of the model were identified:

“Help in offset applications” (1a) “Policy driver – implementation for action, linking areas in development” (2) “Inform env stewardships/BAPs” (2) “Forms targets for conservation action/habitat creation – can predict change” (2)

Potential for the model to be improved to widen its capabilities were also noted:

“Factor in future landscape/land use changes” (1b) “Can use data from other organisations to build up the dataset – prey items etc.” (2)

Other potential advantages were noted in the potential for the model to ensure a consistent level of accuracy and building links between organisations involved in newt conservation:

“Consistency level of accuracy/inaccuracy - NBN gateway [National Biodiversity Network, a national online repository of biological records]” (1b)

“Can build links between different organisations” (2)

Disadvantages

Concerns were raised regarding the quality and sources of the data required to create the model, the implications this would have in terms of limitations, and whether the same model could be applied in all areas:

“It is only as good as the evidence put in” (1a) “Accuracy of the data-uncertainty, professionally obtained” (1b) “Missing data/factor” (1b) “Abundance or pres/abs?” (1b) “Exceptions to the model- concrete pool in the town centre may be imp. but missed,

too much trust in model” (2) “One size doesn’t fit all – multiple models may need to be used” (1b)

There were also concerns about the use and interpretation of the model and whether over-reliance on the model could lead to incorrect decisions being made or the model outputs being rejected by decision makers:

“Planning Authorities understanding and interpretation” (1b) “Can be interpreted differently by different people” (2)

28

Page 29: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Everyone believes it – written well can look impressive – can have missed something” (2)

“Not accepted by statutory authorities – want them on board” (1a)

The issue of the resources required to create and use the model was raised (particularly with regard to small-scale projects), including how the model would be updated:

“May have significant training costs” (1a) “Lack of local resources – councils” (2) “Not good on a small scale project to inform decisions for the effort required” (1a) “How quick in updating the model, considers –time –misuse –obtaining data” (1b) “Review of the model to make more informed management plans” (1b)

Two groups identified factors included within the model as a disadvantage:

“Other priority species – butterflies” (1b) “Chytrid [a fungus that may cause disease in amphibians] – is this mapped, risk but

then other pathways, otter/birds” (2)

Barriers to implementation:

All of the groups identified potential issues with accessibility of the model as a barrier to implementation. This could either be due to a lack of skills of end users, or issues regarding the availability of the software:

“Model too complex to be used by a wide audience (HSI simple)” (1a) “Skill sets – training opportunities – qualified individuals” (1b) “User friendly interface” (1b) “Experience of user – ability” (2) “Standards of modelling – basis” (1b) “Software type, licensed (GIS)/free (MAXENT)” (1b) “Open source/accessibility to models/software” (1b)

The quantity and resolution of the data required to make the model was also considered a barrier to implementation:

“The level of geographical information for an area – might need to collect more data to make the model work” (1a)

“Scale of implementation” (2) “Data informing model – how good is the data?” (2)

One group also noted the ability of the model when compared to field surveys:

“Doesn’t replace on the ground knowledge” (2)

General comments:

A variety of other comments were received. Some elaborated on previously considered topics such as data issues:

“Availability of data – water systems” (1b)

29

Page 30: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Digitised data – some NE data is scanned – maps with drawn points” (1b)

There were concerns over the practical application of the model, in terms of usefulness, physical actions as well as acceptability of the results:

“Can model – but do landowners want to take this up. So actions on the ground difficult” (2)

“Communication” (2) “Could tell you what you already know” (2) “Universality of datasets – say which layers are useful, time – avoid re-inventing the

wheel with volunteer time” (2)

One group identified areas where modelling could be particularly useful:

“Help you think of things which you might not of thought about” (1a) “Be useful for many species, in particular invasive species (INNS)” (1a)

Potential users

The number of groups (out of a total of 12) that identified the different sectors as potential users of modelling is shown in Fig. 4.4. Every group identified ecological consultants and Local Planning Authorities as potential users. Statutory agencies, universities, Local Records Centres, ARGs, national NGOs and local NGOs were all identified as potential users by 11 of the groups. Many groups suggested the model might also be used by ARGs and volunteer surveyors.

NE/

NR

W/S

NH

Loca

l Pla

nnin

g A

utho

rity

Ecol

ogic

al C

onsu

ltan

t

Dev

elop

er

Uni

vers

ity

Land

owne

r

Loca

l Rec

ords

Cen

tre

Vol

unte

er s

urve

yor

AR

G

Nat

iona

l NG

O

Loca

l NG

O

Oth

er (

Uti

liti

es c

ompa

nies

)

Oth

er (

'NIM

BYs

')

Oth

er (

Gov

ernm

ent)

0

2

4

6

8

10

12

Figure 4. The number of groups (out of a maximum of 12) that identified each of the suggested organisations as potential users of predictive modelling.

Recurring themes in participant commentary

Many groups identified the potential of the model to aid great crested newt conservation at a landscape level. A variety of potential applications were noted including strategic land allocation,

30

Page 31: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

assessment of the impacts of development (either large single developments or multiple smaller developments), designing mitigation measures and targeting conservation actions to ensure the greatest benefit could be achieved. The potential of the model to investigate connectivity (either existing situations or proposed conservation measures) was considered a particular advantage.

A number of groups identified the potential for the model to target resources and therefore achieve cost and time savings. They reported that this could be applied to surveying (either for conservation or in relation to development), designing mitigation or targeting conservation effort.

There were a number of concerns regarding the data underpinning the model, principally that data would not be available in sufficient quantity and/or quality in all geographic areas. In some instances this may have been a result of a misunderstanding of the resolution of habitat data required (particularly regarding habitats). However, there was particular concern over the availability of great crested newt distribution data in some areas of the country. Indeed, some participants noted that the examples cited during the introduction were from north-east Wales and Kent, where there is - relatively speaking - a large amount of high quality data available, largely thanks to intensive volunteer effort and agency-funded survey projects.

Additionally several groups raised concerns over the predictive power of the model in relation to limited input data, and questioned whether there would be a ‘ground-truthing’ process. The question of whether the model could and would be updated when new or better data became available was also raised.

Most groups raised the issue of whether the same model would be applicable in different geographical areas or in different habitat types. In some instances this may be attributed to problems related to geographical data absences, however in many cases, respondents were concerned that, for example, a model based on lowland farmland habitats would not work as well in upland or urban habitats, and would lead to erroneous conclusions being drawn.

A number of groups raised issues related to confidence in the model. Firstly, a number of groups were concerned over how well a model would perform and whether its conclusions could be trusted. The possibility that a model would fail to identify situations where great crested newts occurred in unusual habitats was a particular concern. Reservations about the reliability of the model led some groups to suggest that decision makers such as statutory agencies or Local Planning Authorities may not accept the model outputs. Secondly there was a worry that decision makers would not subject the output of the model to sufficient scrutiny (either due to a lack of knowledge of great crested newts, or for political expedience), resulting in inappropriate decisions being made.

Issues regarding the ability of potential end users to access and use the model were raised by many of the groups. There was concern over who would have access to the model, such as whether it would be available to the voluntary sector and whether there would there be a charge to use it or obtain its outputs. A number of groups also raised the issue of ensuring that end users had the knowledge and skills to either run the model (if appropriate) or interpret its output, and the costs associated with training staff. Two groups suggested that a web-based application (such as MAGIC or NBN Gateway) would be the most useful for end-users.

“Straw poll”

31

Page 32: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

The results of the “straw poll” for the final question (“Would you eventually like to see predictive modelling included as a fundamental aspect of great crested newt conservation policy, planning and guidance?”) were:

First session: Yes: 30, No: 5, Don’t know: 9

Second session: Yes: 41, No: 0, Don’t know: 13

Combined: Yes: 71 (73%), No: 5 (5%), Don’t know: 22 (22%); (n= 98).

4.3.3. Targeted end-user workshop

Of the 59 invitees, 24 opted to attend. They represented five sectors as follows: statutory agencies (N=6), local government (N=5), recording groups/NGOs (N=3), consultants (N=6), developers/utilities (N=4).

In the following sections, potential applications identified by the five sector groups have been sorted into general themes for ease of analysis. Participants were asked to identify the specific advantages, disadvantages, implementation requirements and general comments, specific to that application. However, the majority of responses are relevant across more than one application. In many cases the participants indicated it was not clear which specific application an advantage or disadvantage related to (if any at all). Therefore the advantages, disadvantages and implementation requirements are organised into themes that may not relate specifically to individual applications. General comments are included in a separate section at the end.

Every comment received during the workshop has been included below (with the exception of any that were illegible). After each quote, a number in brackets identifies which group the response is from, using these codes:

(1) Developers/Utilities companies

(2) Ecological consultants

(3) Government agencies

(4) Local planning authorities (LPAs)

(5) NGOs and recording groups.

Any emphasis given is that of the participants.

Identifying distribution, range and conservation status:

A number of potential applications were identified within this area, although there was some overlap in them. Developers and utilities companies (and their consultants) saw most potential applications; however this was also of particular interest to recording groups.

“Identifying areas with/without GCNs” (1)

“Identification of high, medium, low risk for newts – ‘risk layer’” (1)

“Identification of areas that are bad for GCNs” (1)

32

Page 33: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Identification of railways, highways, canals as corridors/barriers” (1)

“Options appraisal/scoping (prob. more for large projects)” (2)

“LRC [Local Record Centre]: Input – inspiration for recorders to take on an area; Output – piece of information that can be added to other data when sent [?] out” (5)

“Filling gaps and testing models against factors on the ground” (5)

“Assist with validation of submitted records to LRC’s” (5).

Applications in land-use planning for development:

Most applications within this area were identified by LPAs and others involved in development control (e.g. government agencies). Many of the application identified by developers (see above) could also be considered within this section; however they did not explicitly identify that the application would be in relation to development.

“Options appraisal/scoping (prob. more for large projects)” (2)

“Forward planning and development control” (3)

“Site/land allocations – housing sites, minerals, waste sites etc” (4)

“Development management” (4)

“Spatial planning e.g. site allocation” (4).

Applications in land-use planning for conservation and management:

This area was primarily of interest to government agencies, and to a lesser extent, LPAs and NGOs (as volunteer groups).

“Spatial conservation planning” (3)

“Agri-environment schemes” (3)

“Holistic approach to land management and land use change” (3)

“Green infrastructure” (4)

“Volunteer groups – identifying good areas for surveys and projects” (5).

Applications in survey:

Consultants, government agencies and NGOs identified applications within this area.

“Decreasing survey effort” (2)

“Target survey and monitoring” (3)

“Volunteer groups – identifying good areas for surveys and projects” (5).

Impact assessment and development mitigation:

33

Page 34: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Applications within this area were identified by those involved with development (i.e. developers and their consultants, government agencies involved in protected species licensing, and LPAs).

“Identification of railways, highways, canals as corridors/barriers” (1)

“Impact assessment” (2)

“Focussing mitigation effect possibly including off-setting” (2)

“Strategic planning for compensation” (3)

“Informing derogation licensing/planning/ other consents” (3)

“FCS [Favourable Conservation Status] at relevant geographical scales” (3)

“Development management” (4).

Miscellaneous applications:

“Awareness raising” (4)

“NGO – improve policy and effectiveness of evidence gathering” (5)

Advantages

Note: We consider many of the advantages identified by participants to be applications, and vice versa. However for transparency we are listing the suggestions as reported. Consequently this section should be interpreted in combination with the preceding section.

Forward planning and risk management:

Most advantages in this area were identified by developers, with government agencies and LPAs also seeing some benefits.

“Raise important areas at early in planning” (1)

“Risk model in place might supplement HSI [Habitat Suitability Index] or replace for minor works” (1)

“Mechanism for integrating land use/land use change” (3)

“Manage expectations of GCN works to developers” (4).

Improving efficiency of surveys:

Advantages in improving survey efficiency were identified by all groups (with the exception of developers, who would not be directly involved in surveying).

“Reducing cost, targeting effort” (2)

“Planning tool re survey (cannot survey every pond)” (3)

“Widen certainty over definite need for no surveys, would this really be useful?” (4)

34

Page 35: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Engaging new volunteers to cover high probability areas with no records. More experienced recorders could use it to check presence in low probability areas” (5).

Impact assessment and mitigation:

Use of the model for impact assessment was of particular interest to the LPA group. Government agencies also highlighted an advantage when assessing derogation licences.

“Highlighting fragmentation” (2)

“Context for FCS” (3)

“Improve assessment of schemes including compensation schemes (consider appropriate mitigation)” (3)

“Cumulative impacts” (4)

“Identifying sites with high potential for risk of gcn impacts – as one of many factors to be considered” (4)

“Robust and scientific credibility to decision making, joining up mitigation requirements over developments” (4)

“Better assessment of appropriate areas, costs of development and mitigation” (4)

“Better evidence for decisions back up for expert opinions” (4)

“Useful to consider combined impacts of multiple developments” (4).

Landscape scale conservation and management:

This was of interest to all groups with the exception of the NGOs. However, some of the advantages identified by developers and consultants may be associated with mitigation, rather than pure conservation.

“Better deployment of conservation measures” (1)

“Use of linear structures to improve connectivity” (1)

“Maximising connectivity and more effective use of funds” (2)

“Targeting agri-environment schemes” (3)

“Identify opportunities for habitat gain/connectivity” (4).

Miscellaneous advantages:

NGOs (and to a lesser extent, LPAs) identified a number of advantages in collating and disseminating information regarding range and distribution of great crested newts.

“Info to inform politicians, developers, (public?)” (4)

“Some LRCs have expertise that could run these models. Live updating using recent data from LRC’s” (5)

35

Page 36: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Good for producing thematic maps covering large areas like counties of Natural Areas” (5)

“Simple way of making local groups & recorders useful & also more available for users like planning officers” (5)

“County-wide model useful for many NGOs that are based on them like Wildlife Trusts” (5).

Disadvantages

Reliability and limitations of the model:

There was particular concern about the potential reliability of the model from developers and their consultants, and therefore whether the model would be of practical use in decision making (with respect to risk management, mitigation design, etc.). Government agencies and LPAs were concerned about the reliability (age and extent) of the data underlying the model.

“Risk may be uncertain” (1)

“Level of certainty too risky to influence business decision” (1)

“Extra tier of consultation might increase uncertainty” (1)

“Insufficient data compromises validity” (2)

“[Limitations] in LPA/licensing as well as in house” (2)

“Access to environmental data” (3)

“For development control, scale of models may not be of sufficient resolution” (3)

“Model accuracy about 30%” (3)

“Data sets used for modelling are old or not of sufficient scale” (3)

“Quality/accuracy of records” (4)

“How much more does this give you than local knowledge” (4).

Resources required to develop and use the model:

In relation to concerns over model reliability, developers were apprehensive that the model would be of limited use, i.e. that traditional surveys would still be required and the model would add an increased layer of cost and complexity. Consultants and government agencies were concerned over the costs and staff resources required to use the models, and NGOs over the IT requirements.

“Would not remove the need for surveys/cost” (1)

“Fieldwork still needed” (1)

“Effort in data collation” (2)

“Cost” (2)

“Bigger consultancies have GIS teams – smaller ones would struggle” (2)

36

Page 37: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Need survey and expert job[?]” (3)

“May need multiple models. Resource issues (finance and experts)” (3)

“Resource required to run these models. Expertise and also IT requirements” (5).

Skills and knowledge required to use the model:

Consultants and LPAs were concerned over the knowledge required to use the models. This was also a particular concern for NGOs (volunteer groups and LRCs).

“Need GIS/stats expertise to interpret results + limitations” (2)

“Interpretation by users” (4)

“Most volunteers/groups wouldn’t have the expertise the to run or understand the model” (5)

“Raises more questions than answers (applicable to using models)” (5)

“Some LRCs wouldn’t have the expertise to run models” (5).

Miscellaneous disadvantages:

LPAs identified the potential for the model to be misinterpreted for political reasons. NGOs were concerned over a potential lack of interest.

“NIMBYs misinterpret + use to object against decisions” (4)

“Local recorders might not have long term interest having seen it once” (5).

Implementation

Acceptance of the model:

There was some concern over whether the model would be accepted. In particular the developers and government agencies had reservations over whether each other would accept the output.

“Status of modelling document needs to be enshrined within policy/legislation” (1)

“Still need licence” (1)

“Corporate acceptance of use of model information/conclusions” (3).

Data requirements:

Government agencies were particularly concerned over the need for quality data to inform the model.

“Need for finer resolution data” (3)

“Need for concurrent environmental data” (3).

37

Page 38: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Resourcing:

The resources required to run and interpret the model were a concern of all groups. All groups, particularly the NGOs, identified the lack of expertise as a potential barrier. The LPAs and NGOs also identified a lack of IT equipment as a potential barrier.

“Bigger consultancies have GIS teams – smaller ones would struggle” (2)

“Could ‘buy’ access into service by agreeing to provide data in a specific format” (2)

“Training – technical how to set up + run the model, interpretation and how to communicate to planners and developers” (4)

“Training – technical how to set up + run the model, interpretation and how to communicate to planners and developers” (4)

“ICT resources training for modellers, training for users of maps, training on explanations to developers/public who will provide variables and models for each county/area” (4)

“Hardware ICT available in local authorities” (4)

“Local expertise in GIS needed, IT requirements, training needed. Unlikely that most volunteer groups would have funding to employ someone” (5).

Data flow and dissemination:

Data flow (i.e. ensuring the model has the most up-to-date and precise data) was a concern of all groups. The Government agencies also identified a need to make sure the model outputs were effectively disseminated to interested parties.

“NE [Natural England] already have process in place” (1)

“Improve data collation, curation, dissemination” (1)

“Need central collation of data – not by individual groups” (2)

“Reluctance to give records to consultancies/developers” (2)

“Also not cost-effective/practical to same data for diff[erent] developments” (2)

“Presentation, interpretation and training” (3)

“Turn into plain English/Welsh/Gaelic” (3)

“Integrated roles of public & private sector” (3)

“Ownership – who owns the model?” (4)

“Guidelines & protocols to implement it at a local level to approved standard” (5)

“Collation of eDNA records from across the country – go to companies doing the testing rather than individual consultancies” (5).

General comments

38

Page 39: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Many of the comments within the General Comments section could be applied to other sections, particularly Implementation. A number of comments were submitted on loose sheets of paper, some of which cannot be definitely attributed to a specific group. Where this occurs the group number in brackets is omitted replaced by a question mark.

Overall usefulness of the model:

“Industry needs to feel that mitigation is useful, rather than no benefit to data provided” (1)

“Useful at early stage, but due diligence would mean too risky unless higher certainty” (1)

“Usefulness of models will depend on how it is implemented in legislation” (1)

Apart from county-wide map to ‘start the conversation’, models have limited value to local groups? HSI/survey needed to ground truth pres[ence]-absence” (5).

Additional applications for the model:

“Potential uses for strategic mitigation, land prices – how these could changes as a consequence of modelling” (3)

“Predicted occurrences in new areas can be tested by local volunteers (who like a challenge anyway)” (5)

“Essential for atlas projects, filling gaps” (5)

“Predicted absence can be used in stratification of NARRS [National Amphibian and Reptile Recording Scheme]/other sampling” (5)

“Useful for strategic land acquisition by county WTs [Wildlife Trusts], RSPB, NT, ARC, etc” (5).

Recommendations for implementation:

“NE need to be in a position where they are resourced adequately to process licences, screening process data” (1)

“Would like to see models created and available on-line (whether free or not) + maintained as more data comes in. Then can control presentation of constraints/ limitations + keep it up to date” (2)

“Or could be NE/Defra/LPAs/LRCs etc (could be outsourced by them)” (2)

“Need to link to spatial plans (conservation plans)” (3)

“Other species – e.g. dormice, lesser horseshoe bat, or habitats” (3)

“Resource managed in respect of integrating public and private sector involvement in model” (3)

“Models online” (3)

“Need to look holistically with other ecological matters” (4)

39

Page 40: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

“Local Records Centres to host?” (4)

“A standard ‘Model’ that can be designed/built for LRCs or similar organisations to provide a service much like a “Datasearch’ service from LRC” (4)

“Negative records needed from eDNA for future improvements in modelling to ID variables that might be responsible” (5)

“LRCs can hold local modelling layers, collate with new data” (5)

“LRCs could sell the modelling map as an alert map service” (5)

“UK model – free access via MAGIC” (?)

“Centrally funded” (?).

Other comments:

“Great way of getting local records data widely used and seen, feeling or contributing something useful” (5)

“Great incentive for further recording, seeing data used on a daily basis is gratifying (vs. NARRS – rare feedback)” (5)

“Inspiration for local recorders (to fill map gaps etc)” (5)

“LRCs could manage/develop new iterations of models to include new data (if they have expertise” (5)

“Would you as consultants see this as a ‘threat’?” (4)

[In response to above] “No – welcome anything that can reduce workload, i.e. reduce survey requirements” (2)

“Cost benefit nationally” (?)

“Require GIS on licence and EIA” (?).

Summarising comments by sector

In this section we give a summary of the particular views of each sector, as reported in the groups (detailed comments above) and gleaned from the open discussion sessions.

Developers/Utilities companies. Potential modelling applications included identifying important areas for great crested newts and the development of ‘risk’ layers. Connectivity analysis could also explore the potential of linear habitats such as canals, railways and road verges to act as possible corridors as well as barriers. Advantages included the possibility of early flagging of great crested newt issues in planning, better deployment of conservation resources and supplementing HSI assessments. Disadvantages focused on the fact that modelling would not remove the need for site-specific surveys. Also, the risks and levels of uncertainty associated with model outputs may be too high for due diligence purposes, and may not therefore influence business decisions. In terms of implementation, there was scepticism about the effectiveness of current processes for the

40

Page 41: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

collection, management and dissemination of data, and the purposes to which the data are applied. These issues need to be fixed for industry to have confidence in a modelling approach. There was also concern that adoption of a modelling approach might lead to a further tier of bureaucracy in an already complex licensing process. In response to the question “Does your group think that a modelling approach could help decision-making within your sector?”, three respondents gave a qualified ‘yes’ that predicated upon the issues raised being resolved, and the fourth respondent said ‘no’.

Ecological consultants. Modelling could assist with scoping and appraising options, impact assessment, optimising or decreasing survey effort, and possibly guiding offsetting. The advantages included reducing the costs of targeting survey effort, identifying the impacts of fragmentation and maximising connectivity. However, insufficient data would compromise the validity of models and there would be resource implications in terms of the expertise needed. At present, data are held by different organisations, some of who are reluctant to release data to consultants and developers. Equally, it is not cost-effective to repeatedly request the same data for different developments. A well-managed, centralised, online system for collating data would be needed to ease these issues. Model implementation would be easier for larger companies with existing GIS teams, and smaller operations might struggle, or have to buy-in the expertise. If an independent third party, rather than consultants or developers operated the model, this might improve buy-in to the system, and encourage record submission to improve model performance. The group response to the question “Does your group think that a modelling approach could help decision-making within your sector?” was ‘yes’.

Local government/planning authorities. Modelling could be used to inform development control decisions, spatial planning and site allocation, and to guide green infrastructure networks. It also has the potential to raise awareness of conservation issues and manage expectations of stakeholders. Advantages include the scientific credibility that a modelling process may lend to a project, and the potential to link-up mitigation planning across the wider landscape. On the downside, concerns were expressed over the level of certainty associated with model predictions, and the ‘added value’ that it might provide over and above local knowledge – there was a sense that local surveys would still be needed at a site level. Equally, model outputs could be misused and/or misinterpreted in order to object to decisions. Implementation will require training, expertise and sound communication with developers and planners, clarity concerning ownership and governance of the models, as well as the data that they use. The group response to the question “Does your group think that a modelling approach could help decision-making within your sector?” was ‘yes’.

Government agencies. Potential applications included strategic (forward) planning, optimising agri-environment scheme delivery, targeting surveys, informing licensing decisions, and ensuring a holistic approach to land management and land use change. Advantages include the context that modelling can provide for Favourable Conservation Status and Habitats Directive reporting, and improved assessment for compensation and agri-environment schemes. Modelling could lead to a planning tool for surveys where it is not possible to survey every pond. Disadvantages include the possibility that multiple models may need to be developed for different purposes, resourcing and expertise requirements, and the risk that data used to construct models may be out-of-date or obtained at a scale that is too low-resolution to usefully inform decision-making. Implementation may require ground-truthing and obtaining finer resolution data. There would need to be corporate acceptance of modelling information, more integration between public and private sectors, and

41

Page 42: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

authoritative guidance (provided in all appropriate languages, not just English). Embracing other species and habitats within modelling protocols would be cost-effective. The group response to the question “Does your group think that a modelling approach could help decision-making within your sector?” was ‘yes’.

NGOs and recording groups. Modelling has the potential to identify good areas for newts, and to support the work of volunteer groups and local record centres (LRCs). Advantages include the possibility of validating exiting records and incentivising recorders to fill gaps and carry out ground-truthing. Thematic maps covering counties or natural areas would be useful to the sector. On the downside, most volunteers and many LRCs don’t have the expertise to run models or interpret the outputs. Implementation would therefore require building capacity through training and having guidelines and protocols so that modelling can be carried out to an approved standard at local level. Further standardisation of the collection and processing of eDNA samples may be needed. The group response to the question “Does your group think that a modelling approach could help decision-making within your sector?” was ‘yes’.

4.4. Discussion

4.4.1. Online questionnaire

The results of this questionnaire provide some valuable insights from potential end-users of great crested newt predictive modelling, yet initially it is worth highlighting some methodological constraints. Importantly, almost three quarters of the respondents were from two distinct categories of end-user, statutory agencies (largely Natural England) and Local Planning Authorities. Therefore, we must exercise caution when examining the response rates since those particular end-user groups have certain needs and ways of working, which might not reflect those of other potential end-users. We note an emphasis in the survey results - reinforced by the narratives in free text responses - on the use of models for decisions related to land use and, to a slightly lesser extent, licensing. This is unsurprising given the large caseload relating to land use, but it does risk overshadowing other significant potential uses and users of models, for instance in conservation planning and reporting. Overall we suggest that major results in this survey should be taken primarily to reflect the views of those engaged in decisions relating to land use.

There was an emphasis on needs relating to assessing the potential impacts of projects on great crested newts, mitigation and associated licensing (e.g. Question B1). This was also reflected in the types of information required in current roles (Question B2), with a focus on the data typically required to inform planning decisions and licensing, and relatively little interest in trends. Most data needs were currently met through a combination of bespoke surveys, Local Records Centres, third party sources and professional advice (Question C1). Data were typically needed within one month (Question C2), although the free text responses suggest high variation depending on the context. Surprisingly, there was no clear consensus on whether the seasonal nature or costs of newt survey posed problems for the respondents, with a broad spread of responses. There was a marginal tendency for cost to be considered an issue.

There was a clear interest in exploring the use of predictive models to support great crested newt decision-making. For the question “On the basis of what you know now, how would you rate the potential value of using predictive modelling for your decision-making?” (Question D6) just under half replied “some value”, with approximately the same number of responses split between “very

42

Page 43: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

valuable” and “extremely valuable.” No respondents said there was no value in modelling, whilst only three (8%) said there was “little value”.

In terms of current knowledge, most respondents (66%) either agreed or strongly agreed that they had understanding of predictive models for biodiversity conservation (Question D5). Interestingly, around one quarter of respondents had used predictive models in the past to help with species distribution or status queries, with a range of model types mentioned (Question D1-D4). The experience was generally positive. Nonetheless, around half had experienced some problems with models, and it is worth reflecting on these in order to help with model development in the current project. The issues were: the risk of errors; broad confidence intervals in estimates; queries over model design to take account of ecological issues; risk of individual animals not behaving as a model predicts; the need for expert interpretation; practical implementation issues, especially the cost and time of gathering suitable quality of input data.

Respondents overwhelmingly wanted great crested newt presence/absence data at 1 km square resolution or finer (Question E1), although interestingly the responses did not indicate that the pond level was preferred over coarser scale assessments. The key model outputs preferred by respondents (Question E3) were the ability to predict the following: presence/absence of newts at a particular location; the value to newts of a particular parcel of land; and the likelihood of a particular project impacting on newts, and the scale of such impacts. There was slightly less interest in most other applications mentioned in the survey, though very few respondents said “no importance at all”. The output with least interest was “conservation targets for great crested newts at a national scale”, although this is perhaps unsurprising since most respondents presumably do not work routinely on national issues.

There were unambiguous responses relating to a number of questions on model implementation (Question E2). High average scores indicated end-users find would need reassurance over ease of use and understanding of models, ease of communicating model outputs and demonstrable reliability. It was notable that respondents were divided over whether they would prefer to have access to an output rather than the model itself, but there was a preference for an online, mapped or GIS solution. There was only limited support for access to modelling via a third-party service. Somewhat confusingly, there was a tendency for respondents not to value a cost-benefit analysis of modelling approaches over conventional surveys; this may reflect some ambiguity in the wording of the question as it contradicts a response in Question F1 which reflected a higher consensus for the scale of benefits to be clear. The free-text responses here and in other queries suggest a level of concern over model reliability in practice.

Further operational issues highlight some concern by the respondents (Question F1). It is perhaps unsurprising (given the main user groups responding) that consultees placed great importance on ensuring model outputs would comply with protected species and planning policy and procedure. Respondents also wished to see effective communication of model outputs and their uncertainties, acceptability of the models to other organisations. There were more operational concerns too relating to staff time and expertise.

In conclusion, the consultation has confirmed a substantial interest in seeing predictive modelling explored as a tool in decision-making, especially in relation to land use. There is a clear need for attention to the regulatory and operational landscape in which predictive modelling might operate, as much as the mechanics of modelling itself.

43

Page 44: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

4.4.2. Herpetofauna Workers’ Meeting workshop

In general, workshop participants seem to be broadly receptive to predictive modelling for great crested newts and many positive comments were received, particularly on how it could be used to improve the efficiency of a number of routine conservation and regulatory activities. The fact that the “straw poll” found 73% in favour and only 5% against reinforces this view. However, there was a some scepticism regarding the reliability of the model, especially in in geographic areas with poor data availability. Despite these reservations, participants thought a wide variety of organisations would use the model.

Interestingly, one issue we had anticipated did not get raised explicitly: the potential impact on consultancy income should modelling result in a reduction in mitigation-related fieldwork. One of the workshop leaders (JF) heard these concerns raised informally outside the workshop. However, given that this is a potentially sensitive matter, we suggest an anonymous consultation process would be more effective to tease out whether this is a widely held and genuine concern. See also comments below on this point discussed during the targeted end-user consultation, which indicated that this concern might in fact be a minority view.

We detected that there was a range of prior experience with predictive modelling among the participants. Most, it appeared, had no direct experience. A minority had used models, or had worked in a context where they had been used, sometimes for other species such as bats. The fact that knowledge of modelling was relatively modest among the group is likely to have affected the responses to our proposed approaches.

Based on the comments received from workshop participants, in order to produce a model that would be accepted and used by as wide a range of end users as possible, a number of points should be addressed, as outlined in the following sections.

Providing a clear explanation of what models can do (i.e. what questions can be answered): Many of the comments received may reflect a lack of knowledge about how such models are produced or what they are capable of. Therefore the capabilities of the model need to be clearly conveyed to the end users. For example, the model could predict habitat suitability or likely occupancy. It would not confirm presence or absence, nor predict population size. This should be made clear to end-users so that they do not arrive at false conclusions as a result.

Clearly identifying the limitations of the model and quantifying any uncertainty: Many of the negative comments were concerned with situations where the model might underperform, particularly in failing to predict occupancy in unusual situations or different areas of the country. In order to allay these fears it is suggested that the following suggestions are considered:

Any error in the model should be quantified – e.g. a certainty level could be given as a percentage.

Gaps in the data coverage should be clearly identified. Confidence in the model should be addressed at a local level. For example if a model

based on work in one part of the country is used in another, any differences in performance should be identified to the end user. This could be incorporated into a local scale quantification of error.

44

Page 45: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Any particular situations where the model works less well (e.g. it may predict pond occupancy with lower certainty in particular habitats) should be identified so users can consider this when reviewing the output.

Providing clear guidance on the interpretation of the model output: In order to reduce the risk of misinterpretation of the model outputs, clear guidance should be given on how to interpret them. This should include many of the points discussed in section 4.10, such as what the model actually demonstrates as well as what it does not, and some measure of the reliability of the predictions. This could also include advice on interpreting areas where occupancy is not predicted and the effects of barriers to dispersal.

Allowing access and creating a user-friendly interface: Model implementation routes and methods will be explored in other parts of the current contract, but some insights from the workshop feedback are worth reflecting on here. For example, in order to gain widespread acceptance the model should be as widely available as possible, ideally via a website. If charging is necessary, the possibility of charging different rates for commercial (development related) use as opposed to use for scientific research or conservation. The possibility of making the model available online should be considered, where its user interface would be as simple as possible. One option would be a searchable and ideally clickable layer within existing platforms such as MAGIC or the NBN Gateway.

4.4.3. Targeted end-user workshop

There was consensus across the different sectors that a modelling approach would be useful and could help decision-making processes. All groups highlighted the potential of models to assist with the identification of important areas for great crested newts and in planning conservation and mitigation projects.

Although common themes emerged from the different sectors, there were slight differences in emphasis in terms of both applications and implementation. Those sectors most engaged with development and mitigation expressed concerns that the levels of risk and confidence associated with model predictions may be too wide to usefully inform business decisions. Equally, at site or local levels, modelling would not necessarily reduce the need for on-the-ground species and habitat surveys. Therefore, some participants clearly questioned whether modelling had any value for particular applications where a high level of confidence was required for predictions at a fine spatial scale.

This reinforces the need to critically assess model applications in terms of the consequences of incorrect predictions and associated wider issues of risk-based decision-making, and its foundation in legislative interpretation and regulatory process. Our view is that practitioners currently operate using a range of approaches to decision-making, some of which are based on “expert views”, for which it is impossible to attach confidence levels. Even for traditional field surveys, there is a level of risk; the most important is typically the risk of false negatives. However, there is perhaps something tangible about a field surveys that is easier to grasp: a survey detects newts or it doesn’t. Computer models are, perhaps, more abstract and remote to some, even though any assessment of the presence of newts must come with confidence levels, whether that is via field survey or modelling. It may simply be that by exploring model confidence explicitly, we have drawn attention to risk when previously it was rarely considered.

45

Page 46: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

All groups voiced concerns over current systems for collecting, managing and disseminating great crested newt records and habitat data. Such databases are maintained by a range of organisations, across a range of spatial scales, meaning quality control and validation is variable. Equally, extracting relevant data from multiple systems can be time-consuming and cumbersome. Predictive modelling is unlikely to realise its potential unless the appropriate species and habitat data are readily accessible, and an open access, centralised system would achieve this. Whilst there are currently financial, logistical and other barriers to achieving suitable data exchange, the participants clearly considered the benefits would be considerable.

During the facilitated session with ecological consultants, there was a chance to explore possible concerns over loss of consultancy income (as a consequence of reduced fieldwork due to incorporation of modelling in regulatory processes, as mentioned briefly above in the Herpetofauna Workers’ Meeting workshop). Interestingly, this group overwhelmingly emphasised that potential reduction in fieldwork, and therefore income, was not a concern. Their view was that currently too much survey was undertaken purely to meet regulatory needs but with little actual bearing on conservation or mitigation outcomes. On the contrary, they welcomed modelling since it would likely improve the conservation value of their services. Based on this (admittedly small) sample, it may be that concerns about loss of income are a minority view.

There was a clear view that a trusted party should operate models, government agencies, Local Records Centres and national NGOs were mentioned specifically. There was some debate over whether models should be operated by a body with a local or national remit, with some concern about competing or overlapping models causing confusion. In addition, some participants expressed concern that were developers to start using models, this might raise suspicions over whether they are misused. Participants were also mindful that volunteers (the greatest source of records) might become less willing to submit records if they observe data they collected being misused in models. Underlying these concerns is perhaps the inherent issue of model complexity and, therefore, difficulty for third parties to interpret models. Governance of modelling applications was therefore raised as an important issue to address.

The need for training and building expertise in modelling among the stakeholder groups was widely emphasised. However, increased use of modelling could increase the risk of misinterpretation or misuse of modelling outputs, and enhancing communication between trainers, modellers and decision-makers is needed to minimise this risk.

4.4.4 General conclusions on end-user and stakeholder consultation

The consultation with potential end-users and stakeholders has been thorough, taking into account all key groups and resulting in a clear assessment of their needs. The feedback from these groups has been highly valuable in terms of determining which applications might be in high demand, and how best to ensure modelling is effectively implemented.

A number of general conclusions can be made, taking together the results from all three structured consultation exercises, plus ongoing discussions with potential end-users:

(1) There is a highly positive view of the potential for predictive modelling to assist with decision-making, though with some qualifications and concerns that vary between sectors.

46

Page 47: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

(2) Exploring stakeholder views has been extremely useful for defining potential applications for predictive modelling. This process has refined and added to the original applications suggested during our consultation. The broad areas are:

identifying important areas for great crested newts in a spatially explicit manner status assessment risk assessment impact assessment assessing connectivity informing planning and licensing decisions targeting surveys improving the outcomes and cost-effectiveness of surveys, habitat creation and

management, and mitigation.

Further exploration of these applications is provided in the main Discussion.

(3) Whilst the general concept of predictive modelling is easily explained, or indeed is already understood by end-users, there is a need to raise awareness of its value in particular applications and the attendant limitations.

(4) Potential users frequently raised concerns about using predictive models (although typically within a general acceptance that they have value). The most common concern was over the reliability of model outputs and predictions. There was a sense that models might in fact be redundant if they were not as reliable as traditional field surveys (a view particularly advanced by construction industry representatives in our sample). This in turn relates to understanding of risk and its consequences. Fundamentally, these points relate back to an interpretation of the strict protection provisions; we suggest that a suitably modified interpretation of the legislation could erase the concerns we detected.

(5) A number of operational and communication issues would need to be addressed for predictive modelling to be effective. Much of these relate to standard setting and training. Key points are:

time for staff to learn about using modelling outputs in decision-making; IT capability and capacity for using and interpreting models; ease of communicating model outputs between stakeholders; ease of communicating limitations and uncertainties; interpretation of models to be aided by authoritative, ideally statutory, guidance to

ensure a common approach.

(6) Uptake of modelling would be substantially more likely to be successful if it were available via an easy to use online interface, ideally using a web-GIS approach. The MAGIC system or the NBN Gateway were mentioned as possible candidates.

(7) It is clear that several wider changes would be needed for models to be effective in decision-making, notably:

substantial improvement to great crested newt data collection and data flow;

47

Page 48: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

more explicit integration of risk-based approaches embedded in interpretation of protected species legislation, and associated regulation (notably licensing and land use planning).

(8) There are several mechanisms for use and governance of modelling. Our consultation indicated users would prefer models to be operated by a trusted source, typically a government agency, Local Record Centre or national NGO. There was concern about developers using models, because of the risk of misuse and because this might feedback negatively on data provision by volunteers. The risk of competing models was also raised as a concern. There was no consensus over whether modelling might be best implemented through a local, regional or a central national system. Whatever options are proposed for modelling, there is a need for governance to be clear and authoritative, with common standards.

48

Page 49: sciencesearch.defra.gov.uksciencesearch.defra.gov.uk/Document.aspx?Document=13597... · Web viewIncorporating predictive modelling into great crested newt conservation requires consultation

Annex 1: Online questionnaire structure

Page 1: A brief introduction and background to the questionnaire including a link to the project website

Page 2: A note regarding how the data submitted would be used

Page 3: User details (name, sector, contact details etc.)

Page 4: Questions to identify the users’ current needs for data about great crested newts

Page 5: Questions to identify the users’ current sources of such data

Page 6: Questions to assess the users’ knowledge of predictive modelling

Page 7: Questions to assess the information users would require from a predictive model

Page 8: Questions to assess any potential constraints users may have in using predictive models

Page 9: A request for data to assist in model building

Page 10: Comments

Page 11: End page

49