NatureScot Research Report 1092 - Site Condition Monitoring of golden plover, dunlin and greenshank in Caithness and Sutherland Peatlands Special Protection Area in 2004, 2009, 2015 and 2023

Year of publication: 2024

Authors: Sally Ward (NatureScot)

Cite as: Ward, S. 2024. Site Condition Monitoring of golden plover, dunlin and greenshank in Caithness and Sutherland Peatlands Special Protection Area in 2004, 2009, 2015 and 2023. NatureScot Research Report 1092.

Keywords

wader; golden plover; dunlin; greenshank; density; population; Caithness and Sutherland Peatlands SPA; Flow Country

Background

Caithness and Sutherland Peatlands Special Protection Area (SPA) has internationally important breeding populations of upland birds including golden plover, dunlin and greenshank.

This report presents the results of breeding wader surveys in Caithness and Sutherland Peatlands SPA in 2004, 2009, 2015 and 2023. The same survey plots were counted each year. These surveys are part of NatureScot’s Site Condition Monitoring (SCM) programme. Waders were also counted in some of the same plots in the 1990s, allowing us to assess population trends across 30 years.

Main findings

• Breeding golden plover, dunlin and greenshank were counted in survey plots in Caithness and Sutherland Peatlands SPA in 2004, 2009, 2015 and 2023.
• Golden plover breeding territory density declined by 33% between 2004 and 2009, recovered partially in 2015 (to 24% lower than 2004). There was a further small increase in 2023 (to 21% lower than 2004).
• Dunlin breeding territory density fluctuated between 2004, 2009, 2015 and 2023 with no clear trend across years.
• Greenshank breeding territory density increased consistently between 2004 and 2023, with 51% more occupied territories in 2023 than in 2004.
• 30-year population trends in Caithness and Sutherland Peatlands SPA were consistent with pattern in 2004-2023: the golden plover population has declined, the dunlin population has fluctuated with no consistent long-term trend and the greenshank population has increased. 
• Despite the decline in golden plover and increase in greenshank, golden plover was recorded most frequently and greenshank least frequently in all survey years.

Acknowledgements

We are grateful to the many landowners and land managers who granted access to the survey sites. We also thank the many surveyors who collected and analysed data. In 2004 these were R. B. West, J. S. Kirby, J. Glazebrook, P. Derbyshire, N. Aspey J. Arnold, B. Goodger and C. Stockham of Just Ecology Environmental Consultancy Limited. In 2009 these were P. Bellamy, M. Bruengger, R. Critchlow, S. Holoran, A. Salter, J. Telford, D. Douglas, P. Corbett, M. Eaton, J. Grant, N. Huss, T. Kellett, D. Klein, K. Malone, I. MacLean, N. McLaughlan, S. McLaughlin, L. McNaghten, A. Planting, J. Plowman, D. Shaw, P. Britten, D. Fouracre, W. George and L. Hoffman of the RSPB. In 2015 these were T. Wells, T. Drew, K. Macdonald. K. Liptrot and J. Mach of Atmos Consulting Ltd. In 2023 these were J. B. Bell, D. T. Pointon and T. J. Sykes of Atlantic Ecological Consulting Ltd and R. Boyd, R. Denny of Boyd Ecological Survey. We also thank Robert Hughes (RSPB) for sharing wader data collected in 2023. 

Many NatureScot staff have been involved in planning these surveys and interpreting the results including Diane Baum, Bob Bryson, Rhys Bullman, Kamila Fraser, Simon Cohen, Simon Foster, Sian Haddon, Terry Keatinge, Iain Macdonald, David Mackay, David Patterson, James Plowman, Adam Rose, Karen Reid and Valerie Wilson. Statistical analysis in R was initiated with the help of Thomas Cornulier (then at University of Aberdeen, now at Biomathematics and Statistics Scotland). Megan Towers (NatureScot) updated the statistical analysis to include the 2023 data. Sally Ward (NatureScot) combined the information in the draft 2004-2015 reports with the 2023 data to produce the current report.

Abbreviations 

British Trust for Ornithology (BTO) 

Nature Conservancy Council (NCC)

Royal Society for the Protection of Birds (RSPB)

Scottish Natural Heritage (SNH)

Site of Special Scientific Interest (SSSI)

Site Condition Monitoring (SCM)

Special Protection Area (SPA)

Special Area of Conservation (SAC)

Introduction 

The peatlands of Caithness and Sutherland, also known as the Flow Country, have the largest area of blanket bog in Britain. These peatlands have nationally and internationally important habitats and populations of breeding birds and other species. The importance of these habitats and species is recognised by their designation as 39 Sites of Special Scientific Interest (SSSIs) which together form Caithness and Sutherland Peatlands Special Protection Area (SPA), Special Area of Conservation (SAC) and Ramsar site. 

Between 1979 and 1986, the Nature Conservancy Council (NCC) surveyed birds in 77 monitoring plots across Caithness and Sutherland (Table 1) (Stroud et al., 1987). In 1993-94, Scottish Natural Heritage (SNH) resurveyed 43 of these plots as well as eight new ones (Arnott et al., 1993; Bates et al., 1994). This allowed analysis of changes in wader densities since the surveys in 1979-1986. There had been declines in the three most abundant wader species: dunlin Calidris alpina, golden plover Pluvialis apricaria and greenshank Tringa nebularia (Whitfield, 1997). 

Table 1. A summary of NCC/SNH/NatureScot bird surveys in Caithness and Sutherland. 

¹ 6 of these plots were surveyed in 2009 and 5 of them in 2004, 2015 and 2023 because the plot in Dunbeath Peatlands SSSI was included in the 2009 survey but not in the 2004, 2015 or 2023 surveys.

SNH and NatureScot surveys in 2004, 2009, 2015 and 2023 assessed the condition of the golden plover, dunlin and greenshank features of Caithness and Sutherland Peatlands SPA and component SSSIs as part of the Site Condition Monitoring programme. Twenty to 21 wader survey plots were surveyed across 19 to 20 SSSIs (Tables 1 and 2). A wader survey plot was selected on each component SSSI of Caithness and Sutherland Peatlands SPA that has a breeding bird assemblage feature. The plots thus cover habitat that is particularly suitable for breeding waders.

This report contains the results of the 2004, 2009, 2015 and 2023 wader plot surveys. We also include data from a sub-sample of these plots that were surveyed in the 1990s to assess population trends since 1993. Finally, we consider the wader counts in 1993-2023 in relation to the 1979-1986 counts reported by Stroud et al., 1987.

Methods 

Field survey methods

Golden plover, dunlin and greenshank surveys were carried out in the same monitoring plots in 2004, 2009, 2015 and 2023 other than that the plot with Dunbeath Peatlands SSSI was only surveyed in 2009 and a small part of the southern end of the plot within Rumsdale Peatlands SSSI was omitted in 2015 and 2023 (Tables 1 and 2, Figure 1). The number of survey plots (21) is one more than the number of SSSIs surveyed (20) because Badanloch Bogs SSSI is a relatively small, two-part site with one survey plot in each part. Data are summed across the two Badanloch Bogs plots in all analyses.

Table 2. Wader plots surveyed by SNH/NatureScot in 2004-2023. The plot within Dunbeath Peatlands SSSI was surveyed in 2009 only. The table shows the combined area of the two survey plots within Badanloch Bogs SSSI.

¹ This plot was also surveyed in 1979-1986

² This plot was also surveyed in 1993

³ This plot was also surveyed in 1994

⁴ The area of the Rumsdale Peatlands plot surveyed in 2015 and 2023 was 6.58 ha because the southern tip of the plot was excluded from the survey.

Surveyors used a modified Brown and Shepherd (1993) survey method. Brown and Shepherd recommended that all parts of each survey plot should be within 100 m of the survey route, but logistical considerations meant that a few parts of some plots were up to 200 m from the survey routes in 2004. The 2004 survey route was used in each subsequent survey so that data obtained in different years were collected using comparable survey effort. 

Surveyors stopped at regular intervals to scan and listen for birds whilst walking pre-determined routes through the monitoring plots. They focused on obtaining evidence of breeding territories from adult golden plover, dunlin and greenshank, although any nests or broods were also recorded. Plots were normally surveyed on one day by two surveyors, each covering half the survey route. The surveyors liaised to avoid duplicating records and kept notes about which observations were re-sightings of the same individual when birds moved. They also kept track of which birds were seen simultaneously to ensure that different individuals or pairs were recorded as such. All wader sightings were marked on OS 1:25,000 maps using British Trust for Ornithology (BTO) species and activity codes. Waders were confirmed as breeding if they were observed singing, displaying, in territorial disputes, repeatedly alarm calling, performing distraction displays or if eggs or young were seen. 

All surveys were carried out between 08:30 and 18:00 to avoid peaks in bird activity in the early morning and evening. Surveys were not carried out on days with wind speeds greater than force 5, during heavy prolonged precipitation or when visibility was less than 300 m. If there were occasional squally showers, surveyors took shelter and recommenced surveying when the poor weather had passed. 

The timing of visits varied between years, depending on the weather and surveyor availability. All first visits were made mid-April and late May. Second visits were made between the end of May and early July (Table 3).

Table 3. Wader survey plot visit dates in 2004-2023. 

Minimum pairs estimates of wader populations within survey plots

The Brown and Shepherd method is thought to underestimate breeding territory densities of waders, perhaps by around 14% (Brown and Shepherd, 1993). The breeding density estimates obtained using the Brown and Shepherd method are thus referred to as ‘minimum pairs estimates’. 

We used the same process for estimating populations from field data during each survey as far as possible. Observations of birds not showing territorial behaviour or that were associated with feeding areas rather than breeding territories were removed. Examples of birds not showing breeding territorial behaviour are flocks of golden plover on grassy areas or dunlin and greenshank that flew out of sight in straight lines making no more than a few calls when flushed from loch shores. For each visit, pairs that were seen or heard simultaneously in different places were assessed as being from different territories irrespective of the distance between the observations. The remaining observations for each visit were assumed to relate to independent territories unless the surveyors recorded that birds had moved or observations were close enough together to suggest that two birds were from the same territory. Two birds were counted as being from the same territory if observations were less than 500 m apart for golden plover and greenshank (200 m for dunlin). Where more than one observation was thought to belong to the same territory, the mid-point of the mapped locations was for the summary map unless one of the locations had evidence of nest or chick location, in which case this was used. 

We calculated the total number of territories for each species from the number of territories on the first visit plus any new territories from the second visit. Territories were considered separate between the two visits only if more than 1000 m apart for golden plover and greenshank and 500 m apart for dunlin. These distances reflect the general distances that different species are likely to move during the breeding season (Brown and Shepherd, 1993; Bates et al., 1994).

Estimating golden plover, dunlin and greenshank population change

Statistical analysis was carried out using the mgcv package (version 1.9-0; Wood 2003, 2004, 2011 and 2017; Wood, Pya and Saefken 2016) in R (version 4.2.3; R Core Team 2023). For each species, we assessed whether there was a linear trend in population density over time using a multilevel generalised additive model with Poisson errors and log-link. The response variable was territory count, the explanatory variable was year, and plot was treated as a random intercept. An offset for the log of the plot area was included in the model in order to standardize counts to a unit area, amounting to predicting territory density. Checks of the residuals indicated that the assumptions were met and the fits were adequate.

We used a second model to compare population density between years. This was the same as for the linear trend except that the year was included as a factor rather than a continuous variable. Also, as 1993 and 1994 were two halves of the same survey (only one plot was surveyed in both years) a new factor was created (1993.5) to replace 1993 and 1994. The summary from these models uses t-tests to compare territory density between the year that we have used as a baseline for SCM assessments (2004) and subsequent survey years. The models were fitted on the log-scale, therefore predictions were back-transformed prior to preparing graphs and tables so that they can be interpreted as counts per unit area (densities) and readily compared with previous reports where results were expressed as number of territories km^-2.

Records of other bird species

Surveyors also recorded other birds that were seen within Caithness and Sutherland Peatlands SPA during the course of the wader surveys. The survey work focussed on golden plover, dunlin and greenshank and did not cover the full range of habitats where other species would be expected to be found. Lack of recording a particular species, or lack of finding breeding evidence, therefore does not indicate absence of that species or that it was not breeding.

Results

Golden plover, dunlin and greenshank populations 2004-2023

Minimum pairs estimates of the number of golden plover, dunlin and greenshank territories in each plot in 2004-2023 are given in Tables 4-6. The wader plots varied in size, and the southern tip of the plot within Rumsdale Peatlands SSSI was not surveyed in 2015 or 2023, so we also express the results as territories per km² to facilitate comparison between plots and years (Tables 7-9 and Figures 2-4). Some of the data in Tables 4-9 differ from those presented in the draft 2004 and 2009 reports due to arithmetical corrections since the draft reports were prepared. The figures given in the current report should be used in future.

Table 4. Minimum pairs estimates of the number of golden plover territories in the wader survey plots in 2004-2023. 

Table 5. Minimum pairs estimates of the number of dunlin territories in the wader survey plots in 2004-2023.

Table 6. Minimum pairs estimates of the number of greenshank territories in the wader survey plots in 2004-2023. 

Table 7. Density (number per km²) of golden plover territories in the wader survey plots in 1993-2023, calculated from the data presented in Table 4. 

Table 8. Density (number per km²) of dunlin territories in the wader survey plots in 1993-2023, calculated from the data presented in Table 5. 

Table 9. Density (number per km²) of greenshank territories in the wader survey plots in 1993-2023, calculated from the data presented in Table 6. 

Golden plover was the most frequent of the three species, followed by dunlin. Greenshank was the rarest species (Tables 4-6). Visual assessment of the data shows considerable variation in territory density for all three species both between plots and between years (Tables 4-6 and Figures 2-4). 

Mean golden plover territory density was higher in 2004 than in the three subsequent surveys (Table 7, Figure 2). Mean territory density reached a low point in 2009 when it was 33% less (1.21 pairs km^-2) than in 2004 (1.79 pairs km^-2). There was a partial recovery in 2015 to 1.37 pairs km^-2 (13% more than in 2009 but still 24% less than in 2004). There was a small increase in 2023 to 1.42 pairs km^-2 (18% more than in 2009 but still 21% less than in 2004). 

Mean dunlin territory density fluctuated with no long term trend across survey years: 1.49 pairs km^-2 in 2004, 1.55 pairs km^-2 in 2009 (3% more than in 2004), 1.75 pairs km^-2 in 2015 (17% more than 2004) and 1.57 pairs km^-2 in 2023 (5% more than in 2004) (Table 8, Figure 3). 

Mean greenshank territory density was similar in 2004 (0.72 pairs km^-2) and 2009 (0.71 pairs km^-2) after which there was a consistent increase to 0.79 pairs km^-2 in 2015 (10% more than 2004) and 1.08 pairs km^-2 in 2023 (51% more than 2004) (Table 9, Figure 4). 

Golden plover, dunlin and greenshank populations in 1993-2023

We assessed trends in wader breeding density between 1993 and 2023 by combining the 2004-2023 data with data from 1993 and 1994 surveys that were collected using the Brown and Shepherd method (Arnott et al., 1993; Bates et al., 1994). Eleven of the plots surveyed in 2004-2023 were also surveyed in 1993 or 1994 (Tables 1 and 7-9). Although this data set allows analysis of population trends over a greater time period, there is less detailed information for the earlier years because fewer plots were sampled.

We used a generalised additive model to test whether there were linear trends in wader populations between 1993/4 and 2023 (Annex 1). Golden plover territory density decreased significantly (Figure 5, Annex 1). The modelled decline in golden plover was an average 1.3% per year. This is equivalent to 32.9% decline over the 30 years since the 1993/94 survey. There was no consistent trend in dunlin territory density between 1993/94 and 2023 (Figure 6). Greenshank territory density increased significantly (Figure 7). The modelled increase in greenshank was an average 3.4% per year. This is equivalent to 175% increase over the 30 years since the 1993/94 survey.

We also used a different version of the generalised additive model in combination with t-tests to compare wader populations in 2004 (the baseline year for Site Condition Monitoring assessments) and the other survey years (Annex 2). This is a different way to examine the same data as are shown in Figures 6-8. Golden plover territory density did not vary between 1993/94 and 2004, but was significantly lower in 2009, 2015 and 2023 than in 2004 (Figure 6). Dunlin territory density was slightly greater in 2015 than in 2004 but otherwise did not vary significantly between survey years (Figure 8). Greenshank territory density was lower in 1993/94 than in 2004, did not differ significantly between 2004 and either 2009 or 2015 and was significantly greater in 2023 than in 2004 (Figure 10).

Records of other bird species

Records of other bird species that were seen during the course of the wader survey can be found in Annex 3. 

Discussion

Feature condition and changes in breeding wader populations in Caithness and Sutherland Peatlands SPA in 1993-2023

The changes in golden plover, dunlin and greenshank populations within Caithness and Sutherland Peatlands SPA show the same pattern whether considering statistical significance or condition status in NatureScot’s SCM programme. By statistical significance we mean at least 95% confidence that we are stating correctly whether or not populations have changed between years. By condition status in NatureScot’s SCM programme we mean a loss of more than 25% of the baseline population is the trigger for an unfavourable condition assessment (JNCC 2004). 2004 was used as the baseline year for condition assessments because this was the closest count to when Caithness and Sutherland Peatlands SPA was classified (in 1999). 

Golden plover were the only one of the three waders considered in this report that has been assessed as being in unfavourable condition since Caithness and Sutherland Peatlands SPA was classified. This was in 2009 when territory density was 33% lower than in the baseline year (2004). The golden plover population was also statistically significantly lower in 2009 than in 2004. The golden plover population had recovered just enough by 2015 to bring in back into favourable condition (24% loss from 2004 compared with a 25% loss triggering unfavourable condition). This partial recovery was maintained in 2023 (21% loss compared to 2004). Golden plover territory density was statistically significantly lower in 2015 and 2023 than in 2004, but was within the range of acceptable variation for the golden plover feature to be assessed as being in favourable condition in NatureScot’s SCM programme. Nevertheless, there has been a long term decline in golden plover territory density in the survey plots since the 1993/94 survey, equivalent to an average loss of 1.3% per year or a 32.9% decline over the 30 years since 1993.

There was no consistent change between years in dunlin territory density, with counts being 5-17% more than the 2004 baseline in 2009, 2015 and 2023. This corresponds with the dunlin feature of Caithness and Sutherland Peatlands SPA being assessed as being in ‘favourable, maintained’ condition in all survey years. We found a similar pattern when examining the dunlin population in 1993-2023: fluctuations, but no statistically significant sustained population change. 

Greenshank territory density increased consistently in 2004-2023, being 51% greater in 2023 than 2004. The increase in the greenshank population is also statistically significant, averaging 3.4% per year between 1993/4 and 2023. This is equivalent to 175% increase over the 30 years since 1993. The greenshank feature of Caithness and Sutherland Peatlands SPA was assessed as being in ‘favourable, maintained’ condition in all four survey years. 

Changes in wader populations 1979-2023 

The NCC began surveys of breeding birds in the peatlands of Caithness and Sutherland in 1979, and these continued to 1986 (Stroud et al., 1987). These surveys pre-date classification of Caithness and Sutherland Peatlands SPA in 1999. Some of the 77 plots surveyed were outwith the SPA boundary and others were in places that were not particularly suited to breeding waders. This was because the aim was to survey all breeding birds, to ‘include many examples of all the major peatland habitats’ and ‘to assess typical breeding densities over the whole of the peatlands and not just on the prime sites with exceptional breeding densities’. Most of the plots surveyed in 1993-2023 were also surveyed in 1979-1986 (Tables 1 and 2). We therefore considered whether population trends could be assessed between 1979 and 2023.

The 1979-1986 survey results show that some plots were apparently unsuitable wader habitat as they had zero counts for golden plover, dunlin or greenshank. In contrast, these three wader species bred in all survey plots counted in 1993-2023. The difference in plot suitability for waders would tend to cause lower mean wader densities across the plots surveyed in 1979-1986 than in 2004-2023. 

The 1979-1986 counts were made using a transect survey method which pre-dates development of the Brown and Shepherd (1993) method that was used from 1993 onwards. The 1979-1986 transect counts were based on up to four visits per plot each season (rather than the two visits used in the Brown and Shepherd method). Also, in the absence of field observations that showed which birds were paired with each other, the transect method assumed that single birds were from separate territories if they were more than 50 m apart. The Brown and Shepherd method requires 500 m separation for golden plover and greenshank to be assumed to be from separate territories (200 m for dunlin). Both these differences in survey protocol would be expected to lead to higher counts using the transect method than the Brown and Shepherd method. 

The difficulty in comparing counts made using the transect and Brown and Shepherd survey methods is exemplified by six plots were counted using both methods in 1994 (Bates et al., 1994). Four of these plots were also counted in 2004-2023 (S11, S25, S31 and S40), so these comparisons are more relevant to the data presented in this report than Brown and Shepherd’s assessment that their method underestimates breeding densities of waders by around 14% (Brown and Shepherd, 1993). 

The transect method gave results that ranged from 42% lower to 153% higher than the Brown and Shepherd method for golden plover. The ranges were 53% lower to 500% higher for dunlin and 50% lower to 300% higher for greenshank. These large and variable differences between the results of the two survey methods for individual plots together with the differences in plot selection mean there is no straightforward way to compare wader data collected using the transect method in 1979-1986 with that collected using the Brown and Shepherd method in 1993-2023.

Stroud et al., 1987 reported mean territory densities of 1.76, 2.39 and 0.31 pairs km^-2 for golden plover, dunlin and greenshank in 1979-1986. These are similar to the mean densities recorded in 2004 for golden plover (1.79 pairs km^-2), higher than those recorded for dunlin (1.49 pairs km^-2 in 2004) and substantially lower than those recorded for greenshank 0.72 pairs km^-2 in 2004 rising to 1.08 pairs km^-2 by 2023 (Tables 7-9). As discussed in the paragraphs above, these apparent similarities and differences may be as much to do with changes in survey method as in wader populations. However, these comparisons are consistent with there being a long-term decline in the breeding territory density of golden plover and increase in that of greenshank.

Potential reasons for changes in wader populations

Wader count data can vary for reasons other than long term changes in wader populations. Some of these relate to the countability of birds breeding on the survey plots. Some incubating birds sit tight on their nests are therefore more likely to be missed during counts than birds that flush easily. Some off-duty birds happen to be on the breeding territory during counts whilst others are feeding elsewhere. Birds may happen to nest closer to transects in some years, and therefore be more likely to be flushed from nests. Some birds may start to nest or re-nest earlier or later than others, again changing their countability. 

Interpretation of observations also involves a certain level of judgement to determine which birds are the same ones counted previously, but which have moved, and which ones are new individuals. These differences in bird countability and data interpretation explain some of the year-to-year variation in count data for individual plots. Comparing average territory densities across all plots between years minimises effects of wader countability on overall conclusions.

There are several potential reasons for long-term trends in wader populations in Caithness and Sutherland Peatlands SPA. Differences in trends between species could relate to differences in overwinter survival or suitability of alternative breeding grounds as well as on-site factors. Diseases such as avian influenza also have the potential to affect different species in different ways. There were no obvious reductions in the numbers of golden plover, dunlin or greenshank between 2015 and 2023, suggesting that none of these populations had the sorts of losses due to avian influenza that have been recorded in some other species (Falchieri et al., 2022).

It may be significant that golden plover and dunlin prefer to nest away from plantations (Wilson et al., 2013). Greenshank (the only one of the three target species to increase in numbers) are more tolerant of some scrub, at least in their Scandinavian nesting grounds. The reduction in golden plover counts in 2009 (as plantations near Caithness and Sutherland Peatlands SPA matured) followed by a partial recovery since then (when parts of some plantations closest to the Caithness and Sutherland Peatlands SPA were not replanted after harvesting) is consistent with plantations having an ‘edge effect’ on the golden plover population. Although golden plover counts rose in 2015 and 2023, they have not returned to the 2004 level. This is consistent with there being an ongoing ‘edge effect’ of plantations on golden plover. However, the surveys in this report were not designed to assess edge effect as most of the survey plots are in prime wader habitat well away from plantations. 

Populations of waders breeding within Caithness and Sutherland Peatlands SPA could also be affected by the weather or factors such as disturbance or pollution on their wintering grounds. These could influence the overwinter survival of different species in different ways. Further research would be needed to investigate this. 

Weather during the breeding season is likely to affect wader breeding success too. For example, it was particularly cold at the start of the 2015 survey season, and snow fell in some of the wader plots in April. This could have affected counts in 2015 if territory occupancy was disrupted by birds starting to breed, abandoning nests when it began to snow and not re-establishing breeding territories in time to be counted when survey work began after the snow melted. 

The weather could have affected breeding waders in a different way in 2023, when there was a particularly dry summer. Some pool systems had dried up and the vegetation had become crisp by the time second survey visits were carried out. Very dry ground may have less food for waders. This may have affected territory occupancy during the second survey visit in 2023 if broods moved away from breeding territories to damper areas. Many climate change models suggest that these sorts of extremes in weather may become more common in future, particularly towards the east in the north of Scotland. Any population scale effects on waders breeding in Caithness and Sutherland Peatlands SPA that may be related to climate change will only become apparent in future surveys.

Conclusion

Golden plover breeding territory density in the survey plots in Caithness and Sutherland Peatlands SPA declined by 33% between 2004 and 2009, with a partial recovery in 2015 (to 24% lower than 2004). There was a further small increase in 2023 (to 21% lower than 2004).

Dunlin breeding territory density in the survey plots in Caithness and Sutherland Peatlands SPA fluctuated by 3%, 17% and 5% from the 2004 population in 2009, 2015 and 2023 respectively.

Greenshank breeding territory density in the survey plots in Caithness and Sutherland Peatlands SPA increased consistently, with 51% more occupied territories in 2023 compared with 2004.

30-year population trends in Caithness and Sutherland Peatlands SPA were consistent with pattern in 2004-2023: the golden plover population has declined, the dunlin population has fluctuated with no consistent long term trend and the greenshank population has increased.

Despite the decline in golden plover and increase in greenshank, golden plover was recorded most frequently and greenshank least frequently in all survey years.

References

Arnott, D.A., Whitfield, D.P., Clark, I.R. and Bates, M.A. 1993. A survey of moorland breeding birds in Caithness and Sutherland. A report to Scottish Natural Heritage from Green Associates, Contract No: RASD 93/4-01.

Bates, M.A., Shepherd, K.B., Whitfield, P. and Arnott, D.A. 1994. A breeding wader and upland bird survey of selected sites in Caithness and Sutherland. A report to Scottish Natural Heritage from Green Associates, Contract No: RASD 1994/95 –001.

Brown, A.F. and Shepherd, K.B. 1993. A method for censusing upland breeding waders. Bird Study, 40, 189-195.

Falchieri, M., Reid, S.M., Ross, C.S., James, J., Byrne, A.M.P., Zamfir, M., Brown, I.H., Banyard, A.C., Tyler, G., Philip, E. and Miles, W. 2022. Shift in HPAI infection dynamics causes significant losses in seabird populations across Great Britain. Veterinary Record, 191, 294-296. 

JNCC, 2004. Common Standards Monitoring Guidance for Birds, Version August 2004, JNCC, Peterborough, ISSN 1743-8160.

R Core Team, 2023. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Stroud, D.A., Reed, T.M., Pienkowski, M.W. and Lindsay, R.A. 1987. Birds, bogs and forestry: The peatlands of Caithness and Sutherland. Peterborough, NCC.

Whitfield, D.P. 1997. Waders (Charadrii) on Scotland’s blanket bogs: recent changes in numbers of breeding birds. In: L. Parkyn, R.E. Stoneman and H.A.P. Ingram ed. 1997. Conserving Peatlands. CAP International, Wallingford, pp. 103-111.

Wilson, J.D., Anderson, R., Bailey, S., Jordan, C., Cowie, N.R., Hancock, M.H., Quine, C.P., Russell, N., Stephen, L., Thompson, D.B.A. 2013. Modelling edge effects of mature forest plantations on peatland waders informs landscape-scale conservation. Journal of Applied Ecology, 51,1365-2664.

Wood, S.N. 2003. Thin-plate regression splines. Journal of the Royal Statistical Society (B), 65(1), 95-114.

Wood, S.N. 2004. Stable and efficient multiple smoothing parameter estimation for generalized additive models. Journal of the American Statistical Association, 99, 673-686.

Wood, S.N. 2011. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society (B), 73(1), 3-36

Wood, S.N. 2017. Generalized Additive Models: An Introduction with R. 2^nd edition. Chapman and Hall/CRC.

Wood S.N., Pya N. and Saefken B. 2016. Smoothing parameter and model selection for general smooth models (with discussion). Journal of the American Statistical Association, 111, 1548-1575.

Annexes

Annex 1 - Trends in golden plover, dunlin and greenshank breeding territory density between 1993 and 2023

The tables below show the results of multilevel generalised additive models with Poisson errors and log-link for each species. The response variable is wader territory count, the explanatory variable is year and plot is treated as a random intercept. An offset for the log of the plot area is included in the model in order to standardize counts to a unit area, amounting to predicting territory density.

In each case the model uses a log link function and the formula is wadercount = offset(log(Area)) + as.numeric(Year), random= list(Plot= ~ 1), family= poisson. The R script and data file used to carry out these analyses are given in Annex 4.

Golden plover

The parametric coefficients for golden plover are given in the table below. Adjusted r² = 0.12, n = 89.

We concluded that golden plover territory density declined significantly between 1993/4 and 2023.

Dunlin

The parametric coefficients for dunlin are given in the table below. Adjusted r² = 0.06, n = 89.

We concluded that there was no significant change in dunlin territory density between 1993/4 and 2023.

Greenshank

The parametric coefficients for greenshank are given in the table below. Adjusted r² = 0.29, n = 89.

We concluded that greenshank territory density increased significantly between 1993/4 and 2023.

Annex 2 - Pairwise differences in golden plover, dunlin and greenshank breeding territory density between 2004 and 1993/94, 2009, 2015 and 2023

The tables below show the results of multilevel generalised additive models with Poisson errors and log-link for each species. The response variable is wader territory count, year is included as a factor and plot is treated as a random intercept. An offset for the log of the plot area is included in the model in order to standardize counts to a unit area, amounting to predicting territory density.

In each case the model uses a log link function and the formula is wadercount = offset(log(Area)) + relevel(as.factor(Year), “2004”), random= list(Plot= ~ 1), family= poisson. The R script and data file used to carry out these analyses are given in Annex 4.

Golden plover

The parametric coefficients for golden plover are given in the table below, comparing each other survey with the count in 2004 (the year used to establish the baseline for SCM assessment). Adjusted r² = 0.15, n = 89.

We concluded that golden plover territory density did not differ significantly between 2004 and 1993/94, but was significantly lower in 2009, 2015 and 2023 than in 2004. 

Modelled golden plover density in each survey was as shown below. The fitted mean and 95% confidence intervals are shown in Figure 6.

Dunlin 

The parametric coefficients for dunlin are given in the table below, comparing each other survey with the count in 2004 (the year used to establish the baseline for SCM assessment). Adjusted r² = -0.06, n = 89.

We concluded that dunlin territory density did not differ significantly between 2004 and 1993/94, 2009 and 2023 but was significantly higher in 2015 than in 2004.

Modelled dunlin density in each survey was as shown below. The fitted mean and 95% confidence intervals are shown in Figure 8.

Greenshank 

The parametric coefficients for golden plover are given in the table below, comparing each other survey with the count in 2004 (the year used to establish the baseline for SCM assessment). Adjusted r² = 0.28, n = 89.

We concluded that greenshank territory density was lower in 1993/4 and higher in 2023 than in 2004. Greenshank territory density did not differ significantly between 2004 and 2009 or 2015.

Modelled greenshank density in each survey was as shown below. The fitted mean and 95% confidence intervals are shown in Figure 10.

Annex 3 - Records of other bird species seen during the course of the wader survey 2004-2023

See Excel spreadsheet.

Annex 4 - R script for statistical analyses

The R script and data file used to carry out the analyses presented in this report are available on request from NatureScot.