Public Inquiry into five proposals for wind turbine ...bankssolutions.co.uk/.../08/CON003NRWCURLEW-POE-PEARCE-HIGG… · the impacts of wind farms on upland birds in the UK. Therefore,

Llandinam windfarm Page 1/23 Proof of Evidence of Dr J Pearce-Higgins, British Trust for Ornithology.

Public Inquiry

into five proposals

for wind turbine generating stations and

the

132kV Llandinam connection, known as

Conjoined Wind Farm Inquiry (Powys)

Proof of Evidence

on Curlew in relation to Llandinam

Windfarm

of James Pearce-Higgins BSc (Hons), PhD

on behalf of Natural Resources Wales


1. Introduction

1.1. I am Dr James Pearce-Higgins and since March 2010, have been a principal

ecologist at the British Trust for Ornithology (BTO). This is an independent

charitable research institute combining professional and citizen science aimed

at using evidence of change in wildlife populations, particularly birds, to inform

the public, opinion-formers and environmental policy- and decision-makers.

BTO impartiality enables the data and information to be used both by

Government and NGO campaigners.

1.2. At the BTO I lead on the climate change work which has three main

components; documenting and identifying the impacts of climate change on

biodiversity, undertaking future projections to consider the potential

implications of future climate change and testing and developing approaches

to help conservation adapt to future climate change. I am also responsible for

the research programme associated with the BTO/JNCC/RSPB Breeding Bird

Survey (BBS), which not only informs potential improvements to the survey,

but seeks to develop new modelling techniques to understand the drivers of

population change, and improve predictions of future abundance. This

research programme is largely undertaken by the Population Ecology and

Modelling team that I manage, which comprises five post-doctoral

researchers.

1.3. Prior to my employment at BTO, from March 1999 to 2010, I worked for the

Royal Society for the Protection of Birds (RSPB), running a wide range of

research projects on upland birds. This included work on the impacts of

grazing on upland birds, identifying causes of decline in a range of upland

species, documenting the impacts of climate change and a programme of

research on the impacts of wind farms on birds. Before this, I obtained a PhD

on ‘The ecology of Golden Plovers Pluvialis apricaria in the Peak District.’

from the University of Manchester, and a first-class Honours Zoology degree

from the University of Nottingham.


1.4. I have so far published over 60 peer-reviewed scientific papers, four book

chapters and have just drafted a book on birds and climate change for

publication by Cambridge University Press in 2014. In addition to these core

activities, I am a member of the board of Trustees and Conservation Advisory

Committee of A Rocha UK and of the Scientific Advisory Committee Expert

Panel for Scottish Natural Heritage (SNH). I am an honorary lecturer for the

School of Biological Sciences, University of East Anglia, and associate editor

of the scientific journal, Ibis.

2. Personal experience and background of wind farm research

2.1. The scientific research that I have conducted on wind farms and birds was

largely undertaken when I worked at RSPB, but has continued in my current

role at BTO. This programme of work originated in the mid-2000s when, as a

result of my previous experience into the impacts of recreational disturbance

upon birds, I was increasingly being asked to provide scientific advice to area

and planning staff who were assessing wind farm applications. It soon

became clear that there was a lack of empirical data specifically documenting

the impacts of wind farms on upland birds in the UK. Therefore, I developed

research to fill this gap.

2.2. This research was designed specifically to identify the likely distances over

which birds may be affected by wind farms, and the extent to which wind

farms may result in localised population declines. It was built upon the

methodology of previous work I had led on, documenting the impacts of

disturbance along the Pennine Way long-distance footpath upon breeding

golden plovers (Finney et al. 2005 [CD/CON/003/ORN/001]), which was one

of Biological Conservation’s most cited papers for that year. Given my

previous work documenting the importance of vegetation, topography and

other factors influencing the distribution and abundance of upland birds (e.g.

Pearce-Higgins & Yalden 2004 [CD/CON/003/ORN/004], Pearce-Higgins &

Grant 2006 [CD/CON/003/ORN/005]), it was important that any study looking


at wind farms should attempt to control for these other confounding factors. In

addition, I was keen that we would collect data from as many sites as

possible, in order to maximise the generality of our results.

2.3. Funding for this research was secured from RSPB, Scottish Government and

SNH, and fieldwork was conducted in 2006 and 2007. The main results were

published in Pearce-Higgins et al. (2009 [CD/CON/003/ORN/049]). This paper

was based upon a single year of data from each operational wind farm, and

therefore did not examine whether bird populations changed through time. To

test whether this was the case, there was the need to collate additional

monitoring data from wind farm sites.

2.4. To this end, we secured additional funding from SNH to collate and review

existing monitoring data across a total of 18 wind farms for which comparable

pre- and post-construction monitoring data existed. The aim of this work was

to see whether there was any evidence that bird populations declined at wind

farms during construction or operation. This work was published in Pearce-

Higgins et al. (2012 [CD/CON/003/ORN/050]).

2.5. In addition, given the lack of peer-reviewed published information on changes

in bird populations on wind farms, we published an additional paper

documenting a lack of major change in the distribution and abundance of red

grouse and golden plovers at a single wind farm site monitored 1 and 3 years

after operation (Douglas et al. 2011 [CD/CON/003/ORN/052]).

2.6. More recently, I have led a project at BTO modelling cumulative impacts of

wind farms on birds, funded by the Scottish Wind Bird Steering Group, which

was phase I of a pilot project to develop a potential framework by which

cumulative impacts of wind farms on birds may be robustly and transparently

assessed. Whilst at BTO I have also input to continued research by RSPB to

improve standards of impact assessment and understanding of the impacts of

wind farms on birds (e.g. Douglas et al. 2012 [CD/CON/003/ORN/007]),

assisted BirdLife International with the development of guidance to inform


renewable energy development in areas with migratory soaring birds and

input to a new field study, led by the University of Stirling, to document the

impacts of small wind-turbines on birds and bats (Minderman et al. 2012

[CD/CON/003/ORN/023]).

2.7. To date, I have authored or co-authored eight peer-reviewed scientific

publications on the subject of wind farms and birds. That these are peer-

reviewed means that they have been independently assessed by at least two

other scientists (referees), as well as by the journal editor, and have been

regarded as of sufficient quality for publication. These papers have covered a

wide range of topics, from documenting the impacts of wind farms on birds, to

developing approaches to minimise the risk of detrimental impacts of

developments upon bird populations, and have included analyses that

document a mix of responses by bird populations (not all of the species and

populations studied were negatively impacted).

2.8. Two key papers relevant to this public enquiry (Pearce-Higgins et al. 2009,

2012) were published in the Journal of Applied Ecology, which ‘publishes

novel, high impact papers on the interface between ecological science and

the management of biological resources’. This is one of the highest ranking

ecological journals (ranked 14/129 journals based upon the 2010 ISI impact

factor). Importantly, the 2012 paper was the Editor’s choice in the April 2012

issue in which it was published (Minderman 2012 [CD/CON/003/ORN/006]).

This recognised the key strengths of the paper as being that it explicitly

separated the effects of construction from that of operation, that it included

data from multiple sites, and that it filled a need for studies to demonstrate

population-level impacts of wind turbines on birds.

2.9. This witness statement is largely based upon the science presented in these

two papers, which are amongst the largest-scale studies of displacement of

birds from wind farms anywhere (both in terms of the numbers of sites and

numbers of species covered). Crucially, both have been subject to

independent scientific review prior to publication. Where relevant, I have also


referred to information from other studies. I have therefore presented what I

believe to be the most recent and robust evidence currently available with

which to assess the impacts of wind farms on British upland birds, and

specifically, on curlew.

3. Scope of this evidence

3.1. My evidence will address the impact that the proposed Llandinam Windfarm

Repowering and Extension development, if carried out, would have on

breeding Eurasian Curlew Numenius arquata.

3.2. Specifically, I will address the evidence relating to three areas that underpin

the outstanding objection that Natural Resources Wales (NRW) hold against

the Breeding Birds Protection Plan proposed in relation to the development.

These are that:-

• the buffer around territories should be 800m,

• that uncertainties associated with the mechanisms underpinning the

putative detrimental impacts of wind farm construction on curlew mean

that it cannot be concluded that the disturbance caused by construction

is limited to the turbines alone, and

• that construction activity after 15th February may risk disturbing curlew

returning to the breeding grounds.

4. The evidence that wind farms affect curlew up to 800m

4.1. Here, I shall present a summary of the 2009 and 2012 papers that I have

published on this issue to document the likely impact of wind farms and

curlew, and demonstrate that they may be affected up to 800m from the wind

turbines.


4.2. The ideal methodology of such studies is described by Whitfield et al. (2010

[CD/CON/003/ORN/051]) in technical Appendix 7-B to Llandinam windfarm

repowering and extension ‘Are breeding Eurasian curlew Numenius arquata

displaced by wind energy developments?’

4.3. They write‘…that displacement has been considered the most likely impact of

wind farms on waders (Langston & Pullan 2003) and displacement effects are

best studied with a BACI (Before-After-Control-Impact) design, which involves

survey of bird numbers, distribution or other traits at the wind farm site and a

similar reference site before and after construction… Other research

approaches include the Impact-Reference design where the impact indicators

measured on the assessment (wind farm) area are compared to

measurements from one or more reference sites. In displacement studies the

abundance and/or distribution of birds on the operational wind farm area

compared with those on reference site(s) and any differences related to the

presence of the wind farm, preferable after controlling in analysis for any

confounding environmental factors that may influence abundance. Impact-

gradient designs lack a formal reference site and look for gradients in impact

with distance from the hypothesised impact source – in this case, wind

turbines. In displacement studies, measures of bird abundance would be

expected to decrease with proximity to turbines if an effect occurs. Again,

interpretation is facilitated if potential confounding influences on gradients in

abundance, such as vegetation type, are accounted for. Finally, before-after

designs quantify impact indicators (e.g. bird abundance) on the assessment

(wind farm) area before and after the presumed impact (wind farm

construction) has occurred. A lack of any reference data or replication sorely

limits the ability to interpret such studies objectively, although abrupt changes

after construction can give greater confidence in impact detection…’ Thus,

Whitfield et al. advocate the use of a BACI design (monitoring population

changes through time on wind farm and control sites), but if that is not

possible, suggest that impact-reference design (comparing abundance on

wind farm and control sites) or impact-gradient design (looking at changes in


abundance with increasing distance from the wind turbines) should be used,

preferably after controlling for other confounding factors.

4.4. Of the focal studies relevant to this case, the 2009 study may be regarded as

an impact-gradient with elements of an impact-reference design, that

importantly accounts for confounding influences of vegetation type and other

important environmental factors as advocated by Whitfield et al. (2010). The

2012 study is a BACI-type study. Therefore, the two main studies on which

this evidence is based follow the methodological types advocated by Whitfield

et al. as the best approaches to use when considering the impacts of wind

farms on waders.

5. The Pearce-Higgins et al. 2009 study

5.1. The 2009 study involved surveying 12 wind farm sites in Scotland and

Northern England (the total that we were able to gain access to) in 2006 and

2007. At each site we surveyed a buffer of suitable habitat of up to 1 km from

the turbines on six visits (access restrictions early in the breeding season

limited us to five surveys at three sites), mapping the distribution of birds seen

to a 100m resolution, in order to ‘model associations between wind farm

infrastructure and the distribution of a range of widely distributed upland bird

species’. These data were therefore used to model the probability of a bird

species being sighted on each visit as a function of proximity to the turbines,

on the assumption that were the birds detrimentally affected by the wind farm,

then they would be less-likely to be found close to the turbines than expected

by chance.

5.2. The distribution and abundance of moorland birds is heavily influenced by a

range of other factors, such as topography and vegetation characteristics (e.g.

Pearce-Higgins & Grant 2006), which will vary non-randomly across the

landscape. As wind turbines are also non-randomly located (primarily placed

on hill and ridge-tops), there is the potential for spurious relationships


between proximity to turbines and species’ occurrence to occur as a result of

these confounding factors. For example, a bird species which associates with

low altitude will tend not to be found close to wind turbines located on hill tops

as a result of this altitudinal preference, rather than any actual avoidance of

the turbines. Therefore, it is important that any analysis to look at the effects

of turbines upon bird distribution should account for these factors.

5.3. In order to do this, each wind farm site was divided into squares (cells) of

100x100m dimension up to 500m from the turbines. For logistical reasons,

given the time taken to conduct vegetation surveys within each cell,

200mx200m cells were used beyond this to a 1km radius. Within each cell,

quantitative information about vegetation composition and structure were

recorded in the field, whilst additional information about topography and

proximity to woodland cover were calculated from GIS data. This enabled us

to account for potentially confounding influences on abundance, as

recommended by Whitfield et al. (2010).

5.4. Given the potential difficulties of teasing apart the effects of turbine proximity

upon bird occurrence from these potentially confounding influences, and given

the likely controversial nature of our study, we took a conservative approach

to the analysis. For this, we built upon previous RSPB studies on upland birds

where there was a need to account for potentially confounding effects of other

variables not of interest to the study (confounding variables) before

considering the variables of interest. This approach was first set-out by

Tharme et al. (2001 [CD/CON/003/ORN/020]) looking at the potential benefits

associated with grouse moor management on upland breeding waders and

other species. Given grouse moors only occur on certain habitat types, this

study first narrowed down the range of habitats surveyed, and then modelled

the abundance of moorland birds as a function of habitat and other factors,

before considering the potential additional impacts of grouse moor

management. The validity of this approach was subsequently demonstrated

by experimental evidence, the results of which supported the previous

correlative findings (Fletcher et al. 2010 [CD/CON/003/ORN/030]). This


conservative approach was subsequently used in a study of the effects of

upland grazing on birds (Pearce-Higgins & Grant 2006), before also being

applied in the 2009 study of the distribution of breeding birds around upland

wind farms.

5.5 In this case, we used a statistical model to first predict the occurrence of birds

across each wind farm as a function of vegetation structure and composition,

topography and proximity to woodland cover. Having accounted for these

confounding variables, we additionally considered whether any of the

unexplained variation in where birds were recorded from could be accounted

for by proximity to the turbines. That our approach was conservative is

indicated by the results of a more standard modelling approach that is also

presented in the paper (Appendix S3 of Pearce-Higgins et al. 2009). Here,

when all explanatory variables were considered together, there was evidence

of turbine avoidance by two additional species, skylark and kestrel, which the

conservative analysis had not identified. As the apparent avoidance by curlew

was highlighted in the conservative analysis, this emphasises its likely robust

nature.

5.6. To maximise our ability to separate turbine proximity from other factors likely

to be correlated with turbine proximity, we utilised data from additional non-

wind farm or control sites, which for logistical reasons, were surveyed on

three visits. These were selected ‘to be as similar as possible to the habitat of

the immediate turbine footprint using digital terrain data and satellite images

(Buchanan et al. 2005). Because of time and staffing constraints, non-wind

farm sites (range 180–268 ha) were smaller than wind farms (range 432–932

ha)’.These were not designed to be comparable to the entire wind farm site,

but the area immediately around the turbines, and therefore were similar in

character to the wind farm area, but lacking in turbines, in order to improve

our ability to statistically separate non-wind farm related effects from turbine

proximity effects (Figure 1).


Figure 1. Schematic to illustrate the selection of wind farm and control areas to

minimise the degree of correlation between natural environmental gradients

(such as with topography or vegetation type) and turbine proximity. In this

example, without the control site, all high altitude hill top areas would be close

to turbines.

5.7. As some of these non-wind farm or control areas were under separate land

ownership to the wind farm areas, or may have been managed differently,

there was also the risk that the densities on these control areas may have

differed from those close to the turbines for other reasons. To eliminate this

risk, we conducted two analyses. The first was an impact-gradient design

within the wind farm only, using the 100x100m resolution data to 500m from

the turbines. This is the fine-scale analysis of Pearce-Higgins et al. (2009).

The second was conducted at a 200x200m resolution across the entire wind

farm and control areas (aggregating the 100x100m data as appropriate), and

therefore was more analogous to an impact-reference design, but analysed

taking advantage of the environmental gradients across each wind farm. This

is the large-scale analysis. We therefore used the methods for examining

displacement advocated by Whitfield et al. (2010). We compared the results

of the two approaches to provide greater confidence in the validity of our

conclusions. For the seven species where both analyses were possible, the

same results were identified from both approaches for six. The one exception

was curlew, for reasons described in paragraphs 5.11 and 5.12.

5.8. A number of additional observations presented as part of the 2009 study

further test the validity of our conclusions. Specifically, the frequency of


significant turbine avoidance being detected was much greater than expected

(i.e. avoidance in 10 of 19 tests; Table 2 of Pearce-Higgins et al. 2009). Were

the findings a result of chance, only 1 of these (5 %) would have been

expected to have achieved statistical significance. Secondly, the magnitude

and frequency of avoidance of turbines was much greater than that of tracks

and power-lines. This emphasises turbines as being the feature of a wind

farm that birds appear to avoid the most. Thirdly, our results were robust to

potential pseudo-replication effects as a result of spatial autocorrelation

(spatial autocorrelation implies that locations close to each other may be more

similar than expected by chance because of underlying processes, and if not

taken account of, may lead the data to be treated as having a larger sample

size than is appropriate). Fourthly, the species for which significant effects

were detected did not appear to be those expected were the results purely as

a result of a number of potential statistical biases (Appendix S4 of Pearce-

Higgins et al. 2009).

5.9. Modelled avoidance was described by a mathematical transformation of

distance to follow the expectation that the magnitude of avoidance would be

greatest close to the turbines, and then level-off with increasing distance. This

function produced a smoothed relationship of probability of occurrence with

distance (Figure 3 of Pearce-Higgins et al. 2009) that was used to estimate

the potential reduction in breeding density associated with wind farm

construction (Table 3 of Pearce-Higgins et al. 2009). The precise form of this

function was selected so that there was little effect of variation beyond 1000m,

allowing the data from the control sites to be also included in the large-scale

analysis without those data having undue influence on the final form of the

relationships. It was the statistical significance of this relationship which was

used to test whether species’ showed evidence of turbine avoidance.

5.10. The estimate of the distance over which birds show some avoidance of wind

turbines was provided by examination of the residual probabilities of species’

occurrence in distance bands away from the turbines (Figure 1 of Pearce-


Higgins et al. 2009), after accounting for the potentially confounding effects of

other factors. This figure, for curlew, is replicated in Figure 2.

Figure 2. The probability of occurrence is an estimate of the mean likelihood of

curlew being recorded in each 200x200m cell on any one visit after accounting

for potentially confounding variables (taken from Figure 1 of Pearce-Higgins et

al. (2009) annotated for the purposes of this proof). The error bars indicate the

standard error associated with each estimate; non-overlapping standard errors

can be used to approximate statistically significant differences.

5.11. As outlined above in 5.7, the results of the fine-scale analysis for curlew

showed no significant avoidance of the turbines to 500m, but when repeated

using all data for the large-scale analysis, including the control sites, there

was evidence for significant turbine avoidance. The reason for this apparent

discrepancy is clear upon examination of the residual probabilities of species’

occurrence in distance bands away from the turbines (Figure 2). Curlew show

Limited variation in

occurrence across distance

bands from 0 to 800m

accounting for the lack of fine-

scale avoidance.

Significant contrast

between 600-800m and

800-1000m based on non-

overlapping standard

error bars indicates

avoidance to 800m.

Similar probability of occurrence between

800-1000 m and control sites indicates the

results are not an artefact of the edge of

wind farm sites or control sites being under

different management from wind farm sites.


relatively low levels of occurrence from 0-200m to 600-800m (with limited

variation over the first 500m), and then significantly greater likelihood of

occurrence in the 800-1000m and control sites distance bands. This is the

basis for the 800m avoidance distance for curlew.

5.12. In all other species exhibiting significant avoidance of the turbines, the

avoidance distance was less than 500m, and the results were identified in

both the fine-scale analysis (which extended to only 500m) and large-scale

analyses.

6. The Pearce-Higgins et al. 2012 study

6.1. Although we attempted to control for as many confounding factors as possible

in the Pearce-Higgins et al. (2009) study, it was a correlative study based on

one year of data. No data was presented that showed that populations

actually changed in response to wind farms in the manner expected. For this

reason, we subsequently collated and analysed as much post-construction

monitoring data as possible from upland wind farm sites, to test whether bird

populations actually changed on wind farms during construction or wind farm

operation. This is the work published in Pearce-Higgins et al. (2012). The data

used for this paper were largely derived from post-construction monitoring

data collected by the industry, with some additional data from the 2009 study

used where comparable with pre-construction data. Data were available for a

total of 18 wind farm sites across Britain, of which curlew had been recorded

from 15. Data were also available from 12 reference sites (subsequently

referred to in this proof as control sites).

6.2. Given that different sites were surveyed to different extents around the

turbines, and with different frequencies, there was considerable potential for

error in the data to mask any significant effects. Despite this, across all of the

studies and tests performed, there was much more evidence for statistically


significant changes in bird density on wind farm sites than expected by

chance (11/30 tests), many more than recorded on control sites (2/30 tests).

6.3. Densities of three species, red grouse, curlew and snipe, appeared

significantly reduced on wind farms during construction, although red grouse

populations appeared to recover by the first year of operation. Importantly, for

curlew, this drop in density also contrasted with trends on the control sites,

leading to significantly fewer curlew recorded on the wind farm after

construction than previously, and also compared to densities on the control

sites after construction (Figure 3).

Figure 3. Average curlew densities on wind farms (black bars) and control

sites (white bars) in relation to different periods of wind farm development.

Individual letters link bars that do not differ significantly. Differences between

pairs of bars with all non-matching letters are therefore statistically significant.

The error bars indicate the standard error associated with each estimate.

Pre-construction densities on

wind farm and control sites

comparable.

Curlew densities on wind

farm sites decline during

construction and remain

low post-construction

Curlew densities on control sites remain high

and differ significantly from densities on

wind farm sites post-construction.


6.4. There was no evidence for significant differences in the curlew trends on wind

farm or control sites following operation, although with one caveat that the

post-construction monitoring data analysed in the 2012 paper spanned three

years or fewer at seven sites, which is a relatively small time-series for a long-

lived species. It is therefore possible that some impacts during operation may

not have been identified, or that curlew may show a greater propensity for

subsequent population recovery than these results suggest, although given

the generally declining nature of curlew populations, this latter option currently

seems unlikely.

6.5. There are three important implications of this work. Firstly, the Pearce-Higgins

et al. (2012) study supports the conclusions of Pearce-Higgins et al. (2009).

The species with the greatest magnitude of turbine avoidance from Pearce-

Higgins et al. (2009); curlew – 800m and snipe – 400m, were the species with

the greatest evidence of population decline from Pearce-Higgins et al. (2012).

For species with 200m avoidance or less (golden plover, wheatear and

meadow pipit), there was no evidence of significant population trends on wind

farms, whilst for species with no (red grouse, lapwing, stonechat) or equivocal

(skylark) evidence for avoidance in Pearce-Higgins et al. (2009), populations

on wind farms were either unchanged between pre- and post-construction

(red grouse, lapwing) or increased in abundance during construction

(stonechat, skylark). Thus, qualitatively, the two studies show considerable

support for each other.

6.6. Secondly, specifically, for curlew, the observed magnitude of reduction in

curlew populations as a result of wind farm construction was very similar to

that expected from the models in Pearce-Higgins et al. (2009) that were used

to underpin the 800m avoidance distance. Thus, across all the sites that

contributed to the Pearce-Higgins et al. (2012) study, there was an average

decline in curlew abundance of 36%. This compares to a modelled 42 %

reduction in curlew within an arbitrary 500m buffer around the wind turbines,

or 30% reduction within an arbitrary 1km buffer, from Pearce-Higgins et al.


(2009). Thus, the modelled curlew avoidance from the 2009 paper on which

the 800m avoidance estimate is based, successfully predicts the mean

observed curlew population change that has occurred on constructed wind

farms reported on in 2012.

6.7. Thirdly, Pearce-Higgins et al. (2012) demonstrated that these population

declines appear to occur during construction, rather than wind farm operation,

a result which appeared relatively consistent across the species affected. In

support of this, there was no evidence of more negative curlew population

trends on operational wind farms relative to control areas. Figure 2 of Pearce-

Higgins et al. (2012) shows that modelled curlew population trends on an

operational wind farm were for a 2.9 % per annum increase relative to a 1.4 %

increase on control sites; in other words, no real difference. One potential

caveat to this is that almost half of these sites were monitored for fewer than

three years, which is a relatively short time period over which to assess

population trends in a long-lived wading bird.

6.8. I believe that this evidence points to the main impacts of wind farms on the

birds studied as occurring during the construction period. This potentially

accounts for the apparent lack of changes to the distribution of red grouse and

golden plovers on a wind farm after construction, observed by Douglas et al.

(2011), and apparent lack of variation in curlew nest survival rates with

proximity to turbines reported for Dun Law (Whitfield et al. 2010).

7. Information about the potential mechanisms by which wind farm

construction may disturb curlew

7.1. The precise mechanism underpinning the change in curlew abundance could

not be determined, but Pearce-Higgins et al. (2012) thought the ‘results for

breeding populations suggest that the main negative effects of wind farms

may be through disturbance displacement during construction. High levels of

activity and disturbance are likely to cause birds to vacate territories close to


the turbines, particularly as many upland waders are known to be vulnerable

to disturbance (Finney, Pearce- Higgins & Yalden 2005; Pearce-Higgins et al.

2007). Depending on their subsequent breeding success, they may not return

to breed in subsequent years (Thompson & Hale 1989). The construction of

a barrage has previously been shown to affect the distribution of wintering

waders, including curlew (Burton, Rehfish & Clark 2002), and it is unsurprising

that similar effects apply to breeding birds.’

7.2. Of the upland wader species covered by this work, curlew is probably the

species which is the most sensitive to disturbance, reacting to human

intruders to its breeding territory by alarm calling at a greater distance than

golden plover and dunlin. Yalden & Yalden (1989) [CD/CON/003/ORN/011]

write ‘…the Curlew Numenius arquatus is even more sensitive than the

Golden Plover…it takes flight, alarming loudly, as soon as a human appears

over a distant skyline, at ranges which seem to be around 1km. This

correlates well with the results of Dutch studies which report it to be very

sensitive to human intrusion.’ In addition, analysis of curlew distributions in the

Peak District finds that it was the species which showed the most consistent

negative effects of proximity to footpaths, as a surrogate for disturbance

(Pearce-Higgins et al. 2006 [CD/CON/003/ORN/015]). Although there is no

detailed assessment of precisely what it is about wind farm construction which

detrimentally affects curlew and other bird species, given the sensitivity of

breeding curlews to disturbance by people, my best judgement is that it is the

disturbance associated with construction activity itself, although cannot rule

out the possibility that the novel appearance of turbines in the open landscape

may also contribute to this effect.

7.3. The impact of disturbance associated with the decommissioning of turbines is

likely to be similar to that of wind farm construction.

7.4. The literature on the responsiveness of waders to disturbance suggests that

they are less sensitive to predictable disturbance associated with footpaths

than unpredictable and novel disturbances (Finney et al. 2005). That study


demonstrated that the avoidance of the Pennine Way long-distance footpath

in the Peak District by golden plovers was significantly reduced when the

route of the footpath was resurfaced with flagstones. Although resurfacing

significantly increased the proportion of walkers that remained on the

footpath, the change in bird behaviour was despite a doubling in the number

of people that used it. For this reason, it is possible that the main elements of

disturbance associated with wind farm construction will not be construction

traffic along roads, but the less-predictable, periodically intense construction

activity associated with the turbines themselves and the associated

infrastructure.

7.5. This is a qualitative assessment based upon relatively limited evidence of

which elements of wind farm construction birds are most likely to be sensitive

to. Given this uncertainty, in my opinion it is not possible to exclude

detrimental effects of other potential sources of disturbance to breeding

curlew i.e. it is not possible to conclude that it is only the construction of the

turbines themselves that causes disturbance.

8. The likely timing of curlew arrival to the breeding grounds

8.1. An analysis of nesting dates for curlew provides a number of estimates of the

timing of curlew breeding. Median first egg dates (the date on which curlew will

lay the first of their clutch of eggs) from nest records from 1947-1992 equal 1st

May, although with some evidence that this has advanced (become earlier)

since 1970 (Austin & Crick 1994 [CD/CON/003/ORN/024], Moss et al. 2005

[CD/CON/003/ORN/026]). These encapsulate records from mid-April to early

June. Mid-April laying dates are supported by the results of intensive fieldwork

in Northern Ireland and northern England (Grant et al. 2000

[CD/CON/003/ORN/012]). An additional estimate, based on the back-

calculation of ringing data of chicks is 24th April (Moss et al. 2005). Thus, curlew

appear to lay their eggs from mid-April onwards, with the majority of individuals

having completed egg-laying by early May.


8.2. Pairs return to the breeding grounds to establish territories some time before

egg-laying. Analysis of ringing recoveries shows that ‘Return migration to the

breeding grounds appears to be well under way by March, and is complete by

April.’ (Bainbridge & Minton 1978 [CD/CON/003/ORN/022]). Reviewing this and

other studies, Wernham et al. (2002) [CD/CON/003/ORN/021] conclude that

‘Returning migrants may appear inland as early as late January, and many

southern breeding sites are reoccupied during February; most March recoveries

of British adult breeding birds are close to their natal areas (Bainbridge &

Minton 1978)’.

8.3. Moss et al. (2005) in their review of the timing of breeding of moorland birds,

conclude that many curlew are present on the breeding grounds in late

February, and that the species is present from early March onwards. In support

of this they quote the following additional references in Table 3.4.‘On the moor,

where territories were about 300-400m asl, single curlews sometimes arrived in

mid-February, but the trips, possibly consisting largely of males, usually

returned in the last week in February or in early March (Nethersole-Thompson

1986)’ and ‘Curlew often come and go from the uplands according to vagaries

of the weather during February and March. Heavy snow or hard frost will send

them down to the upper farms, or even back to the low country beyond, there to

reform flocks (Ratcliffe 1990)’.

8.4. Curlew appear to return to their breeding grounds from mid-February onwards,

depending upon the weather. It is likely to be the males which return first, as

there will be a rush to establish or re-establish their breeding territories against

potential competitors. The majority of territories are first occupied between late

February and mid-March. For this reason, there is the risk that construction

activity from mid-February may disturb returning curlew and prevent their

settlement on their usual breeding locations.

8.5. Finally, it is worth briefly considering what may happen to displaced birds.

There is increasing evidence that birds normally show considerable site fidelity,


suggesting there is a strong advantage to remaining in the same location, or

breeding in the same territory. As outlined previously, evidence from wintering

waders suggests that displaced birds perform poorly (Burton et al. 2006

[CD/CON/003/ORN/057]). This is likely to be the case for breeding birds also,

but this has not been tested.

9. Summary of likely impacts of wind farms upon curlew

9.1. I believe there is good evidence, supported by two recent peer-reviewed

published studies (Pearce-Higgins et al. 2009, 2012) using largely separate

sources of data, analysed in completely different ways, both of which were

independently refereed by a high ranking ecological journal, that the presence

of a wind farm is likely to reduce the abundance of breeding curlew at a site.

9.2. The magnitude of that reduction appears relatively consistent between these

two studies and appears to result from fewer curlew occurring within 800m of

the turbines than would otherwise be expected.

9.3. This avoidance appears to occur during wind farm construction. Comparisons

of curlew densities on wind farms with control sites show that declines in

abundance occur during construction. Population trends on constructed wind

farms appear to differ little between wind farm and control sites during

operation. As curlew appear to be highly sensitive to disturbance (more so

than other wader species), and appear to show the greatest sensitivity to wind

farm construction (also more than other wader species), this is likely a result

of direct disturbance associated with the construction activity.

9.4. It is unclear precisely which elements of construction activity may disturb

curlew. Given that waders appear to respond less to predictable disturbance

events than unpredictable ones, it is possible that the unpredictable but

intensive construction activity around turbines and other infrastructure, rather

than increased vehicular traffic, may be responsible, although this has not


been tested. The impact of other locations associated with high disturbance

activity, have not been specifically quantified, but if that disturbance is

unpredictable and intensive in nature, it may also cause disturbance to

breeding birds. It is possible that there may also be some effect of the novel

appearance of vertical turbine structures in an otherwise open landscape, but

this requires further testing.

9.5. Curlew appear to return to their breeding territories from mid-February

onwards, depending upon the weather. Males return first to re-occupy or

establish breeding territories, and at this point, they are likely to be vulnerable

to disturbance that could lead to displacement away from previously occupied

or favoured areas.

9.6. Displaced birds are likely to survive or breed less well than they otherwise

would.

10. Likely implications for a breeding bird protection plan

10. 1 Wind farm construction is likely to cause significant disturbance and

displacement to breeding curlew. This is most likely to be associated with

turbine and other infrastructure construction, but other locations of high and

unpredictable activity may also cause disturbance.

10.2. In order to minimise the risk of disturbance associated with construction,

activity should ideally be avoided in breeding areas during the period when

they will be occupied (likely to be from as early as mid-February to as late as

end-July). In areas of high nest predation or chick mortality, the breeding

season may terminate early with failed breeders likely to have departed by the

latter half of June. In this case, construction during July may be permissible.


10.3. The scale of this disturbance can extend to 800m, based upon the residual

probabilities of curlew occurrence in different distance bands away from

turbines (Figure 2).

10.4. The impact of any restriction of construction activity during the breeding

season should be monitored as a test of whether it successfully reduces the

negative impact of any wind farm upon birds. This should involve as a

minimum, five-visit surveys across the site, following the protocol of Grant et

al. (2000).

Documents

Public Inquiry into five proposals for wind turbine ...bankssolutions.co.uk/.../08/CON003NRWCURLEW-POE-PEARCE-HIGG… · the impacts of wind farms on upland birds in the UK. Therefore,