Abstract
Capsule No evidence for sustained declines in abundance or re‐distribution of two key upland bird species on a wind farm site in the first three years of operation.
Aims To describe changes in the abundance and distribution of birds on an upland wind farm during the first three years of operation.
Methods Surveys to map the distribution of breeding birds were conducted at the wind farm and a nearby control site in 2006 and 2009.
Results Only Willow Ptarmigan (Red Grouse) Lagopus lagopus scotica and European Golden Plover Pluvialis apricaria were sufficiently numerous for analysis. There was no significant difference in the change in abundance of either species between the wind farm and control site, and no evidence that changes in the species' distribution were related to wind farm infrastructure.
Conclusions Upland wind farms may not necessarily result in declines in bird populations in the operational phase. Similar studies across a range of sites should be conducted and published to examine the factors that determine the response of birds to particular developments.
Wind power is a major source of renewable energy generation in the UK, and is set to expand considerably in coming years. Most onshore (land‐based) wind farms in the UK are sited in upland areas because these have a high wind resource, and are remote from areas of high‐density human settlement. Such areas support many habitats and species of high conservation importance (Thompson et al. 1995), which are potentially at risk from the construction of wind farms. Birds are particularly at risk because of mortality from collision with turbines and potential displacement of breeding birds from otherwise suitable habitat (Drewitt & Langston 2006).
A recent study identified significant reductions in the occurrence of seven out of 12 upland bird species within a 500‐m radius of turbines by 15–53% (Pearce‐Higgins et al. 2009). This was based on an analysis of spatial variation in upland bird distribution, rather than monitoring changes in bird populations through time. Such changes in distribution could occur either because of disturbance during construction, from collision mortality, or disturbance during wind farm operation. However, there is a lack of published data on changes in upland bird populations on UK wind farms to examine this issue, with the exception of one study on a pair of Golden Eagles Aquila chrysaetos (Walker et al. 2005).
We attempted to fill this gap by documenting changes in the abundance and distribution of breeding birds at an operational upland wind farm and paired control site. Surveys were first conducted in 2006 (immediately following construction) as part of the wider Pearce‐Higgins et al. (2009) study, and were repeated in 2009 to quantify changes in the distribution and abundance of breeding birds during the first three years of operation. Analyses were conducted to examine changes in the abundance and distribution of two upland bird species that were widespread on the site; Willow Ptarmigan (Red Grouse) Lagopus lagopus scotica and European Golden Plover Pluvialis apricaria (hereafter Golden Plover). The former may be relatively insensitive to wind farm development, while there is evidence of reduced occurrence of Golden Plovers within 200 m of turbines (Pearce‐Higgins et al. 2009). Additionally, Golden Plover are listed on Annex 1 of the EU Birds Directive (Thompson et al. 1995), placing a requirement on developers to avoid adverse effects on protected sites for this species. Specifically, we examined (1) whether the abundance of birds changed between the two periods, and whether that change differed between the wind farm and control site; and (2) whether the distribution of birds could be related to proximity to wind farm infrastructure and whether there was any evidence for that changing between years (either as a result of gains or losses close to the turbines).
Given the lack of published data from such upland wind farms, we present these results in an attempt to stimulate the wider dissemination of the results of monitoring of bird populations on upland wind farms.
METHODS
Study area
The Beinn Tharsuinn study area comprised a wind farm site of 9.3 km2 comprising 17 1.65 megawatt turbines surrounded by a 1‐km buffer, and a control site of 2 km2 located about 2 km to the southwest of the wind farm site. Both sites covered a similar altitudinal range (wind farm mean = 491 m, range 341–670 m asl; control site mean = 587 m, range 447–650 m asl). Both sites were composed of a mosaic of blanket bog and heather‐dominated moorland, with acid grassland and bracken in steep gullies, and had similar mean vegetation height (wind farm mean = 15.1 ± 0.4 cm; control site mean = 16.1 ± 1.4 cm). Construction of wind farm infrastructure was completed prior to the 2006 breeding season, and the site was officially commissioned in September 2006. The level of activity between the two survey periods is likely to have been typical of an operational wind farm site, with a standard schedule of maintenance visits.
Bird surveys
In 2006, six survey visits were made to the wind farm site, and three to the control site (see Pearce‐Higgins et al. [2009] for further details). Four visits to both the wind farm site and control site were made in 2009. The four 2009 visits were timed to match a spread of four of the six visits from 2006 (mean discrepancy = 1.8 days between visits, range = 1.0–3.0) to minimize biases resulting from the varying detectability of birds such as Golden Plovers at different stages of the breeding season (Pearce‐Higgins & Yalden 2005). Accordingly, visits in 2009 were made at approximately 21‐day intervals from late April to late June. The conditions in which surveys were conducted matched standard protocols (Brown & Shepherd 1993), being more than three hours from dawn or dusk, and avoiding periods of strong winds, heavy precipitation and poor visibility. Observers walked transects 200 m apart across each site and plotted the locations of all moorland breeding birds on 1:12500 maps. The only data that have been used for analysis are those from visits that are matched in both years. Thus, the abundance and distribution of birds was summarized from the four repeated visits to the wind farm site, and three repeated visits to the control site.
Because wind farms are located non‐randomly in the landscape, we followed the approach of Pearce‐Higgins et al. (2009), and attempted to take a conservative approach to the analysis by accounting for potentially relevant environmental variables that may also influence bird distribution, prior to investigating the effects of wind farm infrastructure. These environmental variables were detailed measurements of vegetation composition and structure made in the field in 2006, and topographical (gradient and altitude) data, derived from a Digital Terrain Model (all methods described in Pearce‐Higgins et al. [2009]), aggregated at the scale of 200‐m grid squares across both sites (wind farm n = 233 grid squares; control site n = 50). Owing to time constraints vegetation was not recorded in 2009. However any changes in vegetation are most likely to have occurred during construction rather than operation, and we do not consider there to have been large changes in the interval between the two survey periods. The mean distance to the nearest turbine and access track was calculated for each square. These distances were transformed to produce a decreasing rate of decline in value with increasing distance from the turbines, simulating the likely pattern of avoidance (see Equation 1 in Pearce‐Higgins et al. 2009).
Analysis
Analyses were conducted separately for both species in r version 2.9.2 (R Development Core Team 2009).
Changes in abundance
We used chi‐square tests to examine gross changes in abundance on both the wind farm and control sites. For Golden Plovers, two single birds were treated as a breeding pair if separated by less than 200 m, otherwise all registrations of singles or pairs of both species were treated as separate breeding pairs. Following previous studies, we took the maximum count of pairs from survey visits as our measure of abundance (Pearce‐Higgins & Yalden 2005, Pearce‐Higgins & Grant 2006). First, we tested whether there was any evidence for overall population change between 2006 and 2009 by comparing observed abundance in each year against expected values calculated from the mean of the two annual counts per site. Secondly, we examined whether the magnitude of change differed significantly between the wind farm and control site by comparing observed abundance in 2009 with expected values calculated from the mean proportional change in abundance between the two sites.
Changes in distribution
glmms were used to model the distribution of birds across the wind farm and control site. We first tested whether the distribution of birds was significantly affected by wind farm infrastructure, and secondly whether there was any re‐distribution of birds between years in relation to proximity to such infrastructure, while controlling for the effects of relevant environmental predictors. Following Pearce‐Higgins et al. (2009), analyses were conducted at the scale of 200‐m grid squares, employing data from both the wind farm and control site. Bird occurrence was measured as the proportion of visits to a grid square in which each species was recorded, to control for variation in the number of visits to each site. glmms were fitted using the function glmer in package lme4 and contained binomial error structure and logit link, with the identity of each 200‐m grid square (n = 283) specified as a random term, to account for the non‐independence of data from the same squares in the two years.
We employed a two‐stage modelling approach to explain changes in bird distribution that could be attributed to environmental variables, before examining effects of wind farm infrastructure (see Pearce‐Higgins et al. [2009] for a detailed outline and justification of this approach). First, we constructed models of bird distribution using only variables describing vegetation characteristics and topography, incorporating only the most significant term where predictor variables were strongly correlated (r > 0.5). All selected predictors were then entered into a full model along with their interaction with year (two‐level factor) to examine potential changes in bird distribution in relation to habitat. This full model was then simplified by backwards deletion to a minimum adequate model (MAM). Secondly, we tested the significance of wind farm variables (turbine and track proximity), and their interaction with year, when inserted into the MAM; these were entered separately as they were strongly correlated (r = 0.86). Specifically, the wind farm variable alone was the test of significant avoidance of turbine infrastructure, and the interaction between wind farm variable and year, the test for any significant redistribution of birds in relation to wind farm infrastructure. Contrary to Pearce‐Higgins et al. (2009), we did not include an autocovariate term in the analysis because there was no evidence for spatial autocorrelation in the data. The only semi‐variogram model that could be fitted to the residuals from the stage‐one MAM for both species was a null, linear, model with no spatial term (tested using the model fitting module in idrisi 14.0 (Clark Labs 2003).
Because the statistical analysis of change can be problematic (Peach et al. 2001), we used a second alternative approach of ordinal regression to confirm our conclusions. Utilizing data from only those squares in which birds were recorded in either or both of the survey years, squares were classified into a three‐level factor ordered as follows: ‘lost’: recorded in 2006 but not in 2009; ‘stable’: recorded in both years; ‘gained’: recorded in 2009 but not in 2006. This was entered as the response variable in an ordinal regression using the function lrm in the package design. Predictor variables were significant species‐specific environmental variables from the MAMs described earlier, together with wind farm variables (turbine and track proximity), each of which was entered separately with the MAM as previously. Year was not required as an explanatory factor as the form of the response variable inherently accounted for changes between years.
RESULTS
Changes in abundance
There was no significant change in the Red Grouse population (χ2 = 0.12, P = 0.728), or difference in change in abundance between the wind farm and control site (χ2 = 0.06, P = 0.806, Table 1), with densities increasing from an estimated 1.6 to 1.9 pairs km−2 on the wind farm, and remaining constant at 1 pair km−2 on the control site. At a finer‐scale within the wind farm site, this included an increase from six to eight pairs within 500 m of turbines, and from nine to ten pairs at distances greater than 500 m from turbines. Golden Plovers doubled in abundance from an estimated 0.8 pairs km−2 to 1.4 pairs km−2 on the wind farm, and from 1.0 to 3.5 pairs km−2 on the control site (Table 1), although the numbers of birds involved were small enough that this change was not formally statistically significant (χ2 = 3.03, P = 0.082). There was no significant difference in the magnitude of change between sites (χ2 = 1.22, P = 0.269, Table 1). Within the wind farm site, abundance increased from four to nine pairs within 500 m, and from three to four pairs beyond 500 m of turbines.
Table 1. Number of Red Grouse and Golden Plovers recorded at the Beinn Tharsuinn site in 2006 and 2009, summarized as the maximum number of breeding pairs and individuals across survey visits. Wind farm and control sites were surveyed on four and three occasions respectively in each year.
Changes in distribution across the whole site
When using data from all 200‐m grid squares across both the wind farm and control site (n = 283), the minimum adequate model (MAM) of relevant environmental variables for Red Grouse contained a positive correlation with vegetation density only (Table 2). When examining effects of wind farm infrastructure on grouse distribution, there was an additional marginally significant relationship between grouse occurrence and proximity to turbines, and a weaker non‐significant relationship with track proximity (Table 2). Red Grouse, therefore, occurred close to wind farm infrastructure more often than would be expected by chance (Fig. 1). There were no significant interactions between either wind farm variable and year, and, therefore, no significant re‐distribution of birds in relation to wind farm infrastructure during the first three years of operation (Table 2).
Published online:
16 February 2011Figure 1 Percentage of 200‐m grid squares occupied by Red Grouse and Golden Plovers in relation to turbine proximity in 2006 (black bars) and 2009 (white bars). Data are derived from four and three survey visits to the wind farm and control sites respectively in each year. Sample sizes were 283 grid squares across both sites, with 43 squares occupied by Red Grouse in both years, and 23 (in 2006) and 39 (in 2009) occupied by Golden Plovers.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/tbis20/2011/tbis20.v058.i01/00063657.2010.524914/production/images/medium/tbis_a_524914_o_f0001g.gif"})
Figure 1 Percentage of 200‐m grid squares occupied by Red Grouse and Golden Plovers in relation to turbine proximity in 2006 (black bars) and 2009 (white bars). Data are derived from four and three survey visits to the wind farm and control sites respectively in each year. Sample sizes were 283 grid squares across both sites, with 43 squares occupied by Red Grouse in both years, and 23 (in 2006) and 39 (in 2009) occupied by Golden Plovers.
Table 2. glmm outputs examining effect of wind farm operation on the distribution and change in distribution of Red Grouse and Golden Plovers. Modelling was conducted for each bird species separately using a two‐stage approach following Pearce‐Higgins et al. (2009). In the first stage, a minimum adequate model (MAM) of confounding variables describing the presence/absence of birds within 200‐m grid squares was constructed. In the second stage, terms describing potential effects of wind farm infrastructure (testing for interactions with year) were added to the MAM in separate models.
The MAM for Golden Plover based on the same 200‐m data, contained a negative relationship with gradient and positive relationship with altitude (Table 2). Having accounted for these variables, there were no additional significant effects of wind farm infrastructure on Golden Plover distribution (Fig. 1). The distribution of Golden Plovers at Beinn Tharsuinn, therefore, appears unaffected by proximity to turbines or tracks, with no evidence for this lack of association changing through time (Table 2).
Changes in square occupancy
Red Grouse were recorded in 77 of the 200‐m grid squares in either one or both years. Equal numbers of squares (34) were lost and gained between 2006 and 2009, with nine occupied in both years, indicative of a high degree of stochasticity in distribution. Following the previous analysis, neither underlying environmental variables nor turbine infrastructure were significant predictors of change in square occupancy between years (Table 3).
Table 3. Outputs of ordinal regression models examining change in occupancy of 200‐m grid squares expressed as a three‐level ordered response (‘lost’, occupied in 2006 but not in 2009; ‘stable’, occupied in both years; ‘gained’, occupied in 2009 but not in 2006). Predictor variables were species‐specific significant predictors of the intensity of use of grid squares in models examining all squares, with variables describing wind farm infrastructure added separately.
Golden Plovers were recorded in 52 squares in either or both years, with 13 lost, 10 occupied in both years, and 29 gained. As previously, none of the environmental or wind farm infrastructure variables were significantly related to the changes in Golden Plover square occupancy (Table 3).
DISCUSSION
There was little change in the abundance of Red Grouse on both the wind farm and control sites between the two survey years. Golden Plovers doubled in abundance from 2006 to 2009. The apparent breeding abundance of Golden Plovers can vary significantly between years depending upon the co‐incidence of survey visits with the timing of breeding (Pearce‐Higgins & Yalden 2005). However, the magnitude of these changes, plus the fact that they are based on results from four survey visits, suggests they are likely to reflect actual increases in the local population, potentially as a result of weather variation (Yalden & Pearce‐Higgins 1997, Pearce‐Higgins et al. 2010). Weather may affect the breeding abundance of Golden Plovers via several mechanisms, for example a negative correlation with the severity of winter weather (Yalden & Pearce‐Higgins 1997) and a negative correlation with August temperature in previous summers through impacts on the abundance of tipulids, the plovers' main prey (Pearce‐Higgins et al. 2010). Scottish weather data (http://www.metoffice.gov.uk/climate/uk/datasets/Tmean/ranked/Scotland.txt) show that mean winter temperature in the intervening three winters was 0.3°C warmer than the mean for the previous ten years, while August temperature from 2005–2007 (because of the two‐year lag) was 0.9°C colder than the previous ten‐year mean. Thus, weather conditions between the two surveys are likely to have benefited the regional Golden Plover population and may at least partly explain the observed population increase. Importantly, there was no evidence for a sustained decline in abundance of either species during the three years of wind farm operation studied. Further, there was no difference in trend between the wind farm and control site, although the small size of the control site may have limited our ability to identify such a contrast.
The analysis of species distribution highlighted a positive association between Red Grouse occurrence and turbine proximity. Given the high correlation between turbine and track proximity, this reflects the previous results of Pearce‐Higgins et al. (2009) that Red Grouse show a general association with tracks. Anecdotal evidence suggests Red Grouse may use tracks as a source of grit, which they ingest to aid digestion (Watson & Moss 2008), which may account for the slightly stronger association with turbines, which are associated with a greater surface area of road. Additionally, cocks may display along linear features, such as tracks, to demark territory boundaries, and therefore may be more detectable close to such features (Watson & Moss 2008). Importantly there was no significant change in the relationships between grouse occurrence and either turbine or track proximity between 2006 and 2009 and, therefore, no evidence for re‐distribution in Red Grouse in response to wind farm operation.
There was no evidence for any significant avoidance of the turbines or tracks by Golden Plovers at the site. This contrasts with the wider results of Pearce‐Higgins et al. (2009), who found reduced occurrence of Golden Plovers within 200 m of turbines across 12 upland wind farms and whose analysis included 2006 data from the current site. This suggests that under some circumstances, Golden Plovers may be more tolerant of wind farm infrastructure than the generic results of Pearce‐Higgins et al. (2009) suggest. More data on temporal changes in distribution from a range of sites are required to determine what may cause such variation, but it may be related to the distribution of suitable habitat relative to the location of the turbines. Further, it appears that the levels of disturbance associated with the tracks on the wind farm during operation were insufficient to affect Golden Plover distribution (Finney et al. 2005, Pearce‐Higgins et al. 2007). As with Red Grouse, there was no evidence for changes in Golden Plover distribution between years in relation to wind farm variables.
There are clearly limits to the inferences that can be made from a single site with two years of survey data and modest bird densities, which may have limited the power of our analyses. The small extent of the control site (initially selected for use in the Pearce‐Higgins et al. 2009 study) may have imposed an additional limitation, although given that both Red Grouse and Golden Plovers increased in abundance on the wind farm site, this did not significantly affect our results. Therefore, there was no evidence for sustained negative impacts of the Beinn Tharsuinn wind farm upon local Red Grouse and Golden Plover populations during the first three years of operation, with neither a decline in abundance nor a re‐distribution of birds away from the tracks or turbines.
Do these results conflict with the recent work of Pearce‐Higgins et al. (2009)? Not necessarily. First, we have only examined changes in the abundance and distribution of birds following operation, rather than comparing abundance and distribution before, during and after construction. Given the potential sensitivity of upland birds to disturbance (Finney et al. 2005, Pearce‐Higgins et al. 2007), it is possible that effects during the construction phase may be more severe than those of wind farm operation, and further studies should be conducted to test this. Second, the results of a single study location cannot be used to support or refute the more general results of Pearce‐Higgins et al. (2009), which is why we would encourage others to publish the results of similar studies from other sites, to facilitate a wider analysis of such data. Ideally, post‐construction monitoring should occur in years 1, 2, 3, 5, 10 and 15 following wind farm operation (Scottish Natural Heritage 2009). This would account for potential fluctuations in populations between years and allow an assessment of whether birds re‐occupy sites following construction, and any lag effects associated with this.
This study provides a model for such work, although others should aim for similar‐sized control sites to the wind farm survey area to increase statistical power. Data should ideally be collected using a minimum of four visits (cf. Calladine et al. 2009) to assess appropriately distribution and abundance and to control for underlying environmental influences on bird distributions, such as vegetation and topography.
| Wind farm | Control | ||||
|---|---|---|---|---|---|
| Species | Year | Max. pairs | Max. individuals | Max. pairs | Max. individuals |
| Red Grouse | 2006 | 15 | 16 | 2 | 2 |
| 2009 | 18 | 18 | 2 | 3 | |
| Golden Plover | 2006 | 7 | 10 | 2 | 4 |
| 2009 | 13 | 21 | 7 | 9 | |
| Term | Slope | 95% CI | df | χ2 | P |
|---|---|---|---|---|---|
| Red Grouse | |||||
| MAM | |||||
| Vegetation density | 0.462 | 0.007–0.917 | 1 | 4.87 | 0.027 |
| Wind farm variables added to MAM | |||||
| Turbine proximity | 1.168 | −0.048–2.384 | 1 | 3.84 | 0.050 |
| Turbine proximity × Year | 1 | 0.92 | 0.338 | ||
| Track proximity | 0.773 | −0.159–1.705 | 1 | 2.84 | 0.092 |
| Track proximity × Year | 1 | 0.87 | 0.352 | ||
| Golden Plover | |||||
| MAM | |||||
| Gradient | −0.302 | −0.214–0.390 | 1 | 61.26 | < 0.001 |
| Altitude (/100) | 0.845 | 0.452–1.239 | 1 | 17.57 | < 0.001 |
| Year | 1 | 4.93 | 0.026 | ||
| Wind farm variables added to MAM | |||||
| Turbine proximity | 0.375 | −1.327–2.077 | 1 | 0.44 | 0.509 |
| Turbine proximity × Year | 1 | 0.01 | 0.970 | ||
| Track proximity | 0.099 | −1.337–1.535 | 1 | 0.01 | 0.919 |
| Track proximity × Year | 1 | 0.01 | 0.929 |
| Term | Slope | 95% CI | df | χ2 | P |
|---|---|---|---|---|---|
| Red Grouse (n: lost = 34, stable = 9, gained = 34) | |||||
| Model 1: Turbine proximity | |||||
| Vegetation density | 0.485 | −0.771–1.740 | 1 | 0.77 | 0.442 |
| Turbine proximity | 0.226 | −2.248–2.700 | 1 | 0.18 | 0.856 |
| Model 2: Track proximity | |||||
| Vegetation density | 0.607 | −0.498–1.711 | 1 | 1.09 | 0.274 |
| Track proximity | −0.128 | −1.903–1.647 | 1 | 0.14 | 0.886 |
| Golden Plover (n: lost = 13, stable = 10, gained = 29) | |||||
| Model 1: Turbine proximity | |||||
| Gradient | 0.174 | −0.060–0.407 | 1 | 1.49 | 0.135 |
| Altitude (/100) | −0.608 | −1.759–0.542 | 1 | 1.06 | 0.288 |
| Turbine proximity | −0.015 | −2.934–2.904 | 1 | 0.01 | 0.992 |
| Model 2: Track proximity | |||||
| Gradient | 0.169 | −0.060–0.398 | 1 | 1.48 | 0.139 |
| Altitude (/100) | −0.629 | −1.751–0.493 | 1 | 1.13 | 0.260 |
| Track proximity | −0.112 | −2.481–2.257 | 1 | 0.09 | 0.924 |
ACKNOWLEDGEMENTS
Fieldwork in 2006 was funded by RSPB, the Scottish Government and SNH. The repeat survey in 2009 and this analysis were funded by Scottish Power Renewables. We are particularly grateful to David McArthur and Peter Robson for their support and encouragement. Adam Seward and Ewan Munro conducted bird and vegetation surveys in 2006, and Rob Critchlow, Susan Holoran and Paul Bellamy conducted bird surveys in 2009. We thank Jenny Gill, an anonymous referee, Rowena Langston, Jeremy Wilson and Benedict Grove for useful comments on an earlier draft.
REFERENCES
- Brown, A.F. and Shepherd, K.B. 1993. A method for censusing upland breeding waders. Bird Study, 40: 189–195. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
- Calladine, J., Garner, G., Wernham, C. and Thiel, A. 2009. The influence of survey frequency on population estimates of moorland breeding birds. Bird Study, 56: 381–388. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
- Clark Labs. 2003. IDRISI 14.0, Worcester, MA: Clark Labs. [Google Scholar]
- Drewitt, A.L. and Langston, R.H.W. 2006. Assessing the impacts of wind farms on birds. Ibis, 148: 29–42. [Crossref], [Web of Science ®], [Google Scholar]
- Finney, S.K., Pearce‐Higgins, J.W. and Yalden, D.W. 2005. The effect of recreational disturbance on an upland breeding bird, the golden plover Pluvialis apricaria. Biol. Conserv., 121: 53–63. [Crossref], [Google Scholar]
- Peach, W.J., Lovett, L.J., Wotton, S.R. and Jeffs, C. 2001. Countryside stewardship delivers cirl buntings (Emberiza cirlus) in Devon, UK. Biol. Conserv., 101: 361–373. [Crossref], [Web of Science ®], [Google Scholar]
- Pearce‐Higgins, J.W. and Grant, M.C. 2006. Relationships between bird abundance and the composition and structure of moorland vegetation. Bird Study, 53: 112–125. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
- Pearce‐Higgins, J.W. and Yalden, D.W. 2005. Difficulties of counting breeding Golden Plovers Pluvialis apricaria. Bird Study, 52: 339–342. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
- Pearce‐Higgins, J.W., Finney, S.K., Yalden, D.W. and Langston, R.H.W. 2007. Testing the effects of recreational disturbance on two upland breeding waders. Ibis, 149: 45–55. [Crossref], [Google Scholar]
- Pearce‐Higgins, J.W., Stephen, L., Langston, R.H.W., Bainbridge, I.P. and Bullman, R. 2009. The distribution of breeding birds around upland wind farms. J. Appl. Ecol., 46: 1323–1331. [Web of Science ®], [Google Scholar]
- Pearce‐Higgins, J.W., Dennis, P., Whittingham, M.J. and Yalden, D.W. 2010. Impacts of climate on prey abundance account for fluctuations in a population of a northern wader at the southern edge of its range. Global Change Biol., 16: 12–23. [Crossref], [Web of Science ®], [Google Scholar]
- R Development Core Team. 2009. R: A language and environment for statistical computing Available at: http://www.R-project.org (accessed 16 February 2010) [Google Scholar]
- Scottish Natural Heritage. 2009. “Guidance on methods for monitoring bird populations at onshore windfarms”. SNH Guidance Note [Google Scholar]
- Thompson, D.B.A., MacDonald, A.J., Marsden, J.H. and Galbraith, C.A. 1995. Upland heather moorland in Great Britain: a review of international importance, vegetation change and some objectives for nature conservation. Biol. Conserv., 71: 163–178. [Crossref], [Web of Science ®], [Google Scholar]
- Walker, D., McGrady, M., McCluskie, A., Madders, M. and McLeod, D.R.A. 2005. Resident Golden Eagle ranging behaviour before and after construction of a windfarm in Argyll. Scot. Birds, 25: 24–40. [Google Scholar]
- Watson, A. and Moss, R. 2008. Grouse, Glasgow, , UK: Collins. [Google Scholar]
- Yalden, D.W. and Pearce‐Higgins, J.W. 1997. Density‐dependence and winter weather as factors affecting the size of a population of Golden Plovers Pluvialis apricaria. Bird Study, 44: 227–234. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]