POPULATION TRENDS OF BREEDING SEABIRDS COLONIES IN SCOTTISH SPAS

2 Methods

Data were obtained from the JNCC's Seabird Monitoring Programme ( SMP), SNH, bird observatories such as Fair Isle and Isle of May, and any other regularly manned and monitored reserves, such as Scottish Wildlife Trusts ( SWT) Handa Island reserve, or the Royal Society for the Protection of Birds ( RSPB) reserve at Marwick Head. These data were then collated together into a single Excel spreadsheet, where they could be arranged in order of SPA, species, and year. If there were multiple counts for the same count site in the same year, these were considered to be partial counts and were summed.

There were several issues identified with collating the data available. The SMP began before the Birds Directive and the need for SPA designation, and consequently the count areas and sites used are not necessarily compatible with the boundaries of relevant SPAs.

Therefore, the majority of count sites were selected manually, by eye, by the JNCC's SMP coordinator prior to the production of the Work Package E report. For a small number of SPAs this was deemed an unsuitable approach (Fetlar SPA, Hoy SPA, Papa Stour SPA, Papa Westray SPA, Rousay SPA, and South Uist Machair and Lochs SPA). It was therefore agreed with JNCC (Roddy Mavor) to apply a GIS-based approach to isolate count sites within these SPAs, using a spatial join tool. The spatial join tool overlays the count sites with the SPA shapefiles, thus isolating all positions that fall within the boundaries of an SPA. The count sites observed during data collection are presented in Table 3.1 alongside the Ordnance Survey grid references.

Where necessary some incompatible counts (due to count type) were eliminated. The data presented here represents the largest dataset of compatible counts for each species at each SPA. Most of the variation in count types pertains to different methods of counting or expressing breeding 'units'. Therefore, the following count types were regarded as breeding 'units' and were considered compatible for incorporation into the current review. They are presented in the following species accounts as 'pairs':

• Occupied nests;
• Occupied sites;
• Occupied burrows;
• Occupied territories.

Counts for auks were predominantly of individuals on land apart from Atlantic puffins, which were counted in occupied burrows (counts for individuals on land have been used as a proxy for occupied burrows in some innaccessable sites, meaning that Hermaness and Mingulay and Berneray include data of both types).

In this report, we provide an overview of data availability and population trends at individual SPAs on a species by species basis in the form of species accounts. An index of 'data availability' has been tabulated within each species account to assess the temporal and (for SPAs with multiple count sites) the geographical coverage at each SPA. The index in no way comments on the reliability of the data held for any SPA. The number of years since the first Seabird 2000 count for each species (2000 or the nearest year with a complete count between 1998 and 2002) where all count sites at an SPA were observed is presented. The first Seabird 2000 count is defined as the first year where all colony counts were observed between 1998 and 2002. A proportion indicating the number of years surveyed against the total number of possible survey years (up until 2011) since the first Seabird 2000 count is presented to provide an indication of the temporal coverage of full SPA colony counts. The number of years during which only part of an SPA was observed is also tabulated, and in this case the proportion represents the number of counts observed as a percentage of the total number of possible counts, i.e. every count site recorded every year within the SPA. This proportion provides an indication of the geographic coverage for each species at each SPA. Where values are low it is likely that application of the data within marine renewable EIAs and HRAs will be accompanied by increased uncertainty in the final risk determinations.

For species where there is at least five complete years of data (counted at every count site within an SPA) since the first Seabird 2000 count population trends for that colony have been produced. Where there is a variation from this rule it is detailed within each species account. The raw counts are presented graphically with a polynomial smooth trend line and associated standard errors around the fitted values produced using the statistics program R.

As might be expected, fluctuations in local population sizes over time suggest that a combination of different variables affects the abundance of birds counted at any given site each year. As a result, whilst in many cases it may be possible to detect general population trends, confidence will be low for a model with year as the sole explanatory variable. The graphs produced in this report show that confidence intervals around predictions into the future based on such a model are so great that these predictions will provide no more value and could prove less accurate for assessing the impacts of marine renewable in the context of EIA and HRA than simply using the latest available count. We therefore recommend that consenting decisions should be based on field collected data rather than attempting to extrapolate from existing data.