Scottish Marine and Freshwater Science Volume 6 Number 4: A Mapping Study of the Overlap of Potential Eurasian Beaver (Castor fiber) Habitat and Atlantic Salmon (Salmo salar) Distribution in Scotland

2. Methods

GIS was used to determine the percentage of river habitat known to be currently occupied by Atlantic salmon that would be predicted to be used by beavers should they extend to their full potential range. Salmon distribution data, provided by Marine Scotland Science ( MSS) and last updated in 2008, includes the spatial range over which juvenile salmon have been detected in Scotland. Potential beaver habitat distribution data, produced, using the methodology of Webb et al. (1997), and provided by Scottish Natural Heritage ( SNH), includes habitat types and topographies known to be used by the mammals in other European countries. The data set includes a wide range of habitat types, including small groups of trees and scrubland, including willow and alder. Hence, it emphasises the widest area of habitat likely to be used by beavers rather than only those areas that might sustain territories over long periods of time and hence encourage production of dams.

2.1 Datasets

Salmon distribution was compiled initially by Gardiner & Egglishaw (1986) and updated in 2008 with more recent data from surveys conducted by Marine Scotland Science and Fisheries Trusts. The updated salmon distribution data were incorporated in the Centre for Ecology and Hydrology ( CEH) 1:50,000 Digital River Network ( DRN) as a Scotland-wide line dataset. This information was transferred to Ordnance Survey MasterMap ( OSMM) to create area-based polygon data for each catchment and subsequently split twice. Firstly to separate lochs and rivers, locations were identified using Scottish Environment Protection Agency ( SEPA) loch coding in the CEH DRN and Ordnance Survey 1;10,000 data as a guide. Secondly, river polygons were split based on SEPA 1:50,000 stream width data, into two river types: those areas below and equal to, and those above 10m in width.

Potential beaver habitat distribution data were produced by SNH (August 2011) from fresh water, woodland and Ordnance Survey datasets as listed below, using a methodology based on earlier studies (Webb et al, 1997). Suitable beaver woodland was compiled from areas with existing broadleaved and mixed broadleaved woodland (Highland Birchwood's Scottish Semi-Natural Woodland Inventory - SSNWI - and Forestry Commission's National Inventory for Woodlands and Trees - NIWT), lying within 200m of suitable water or wetland habitat. Tidal areas were excluded as were areas above 400m, and stream reaches of greater than 2% gradient using a slope analysis of OS 1:50,000 Panorama Digital Terrain Model ( DTM) data.

Catchment boundaries for the Rivers Awe, Ayr, Conon, North Esk, Tay and Tweed were extracted from data produced by SEPA from CEH 1:50,000 DTM data. For the study, the Tay catchment excluded the River Earn and the Tweed catchment included the River Whiteadder. Since SNH potential beaver habitat datasets did not extend into England, the Tweed catchment boundary was restricted to the area within Scotland, using OS Panorama 1:50,000 data at the Scottish Border.

2.2 Salmon and beaver distribution overlap

Salmon rivers data were required in polygon format to quantify the area of potential overlap. The salmon rivers DRN (line) dataset contains no area information, and processing used salmon rivers OSMM (polygon) data, extended with OSMM line data where necessary. With the potential beaver habitat datasets restricted to areas below 400m in elevation, salmon distribution polygons were checked against OS Profile 1:10,000 contour lines and salmon river sections above 400m altitude, which occurred only on the North Esk and Tay, were excluded.

To calculate the area of salmon rivers in proximity to potential beaver habitat at a catchment scale, suitable beaver woodland was "buffered" by 50m and the total area and percentage of salmon rivers within this buffer zone was calculated. Secondly, using the salmon river type categories (section 2.1), this overlap methodology was repeated to provide area values and percentages specific to each river category.

The method of analysis limited overlap to a 50m buffer zone, which was the smallest scale applicable given data resolution. Salmon lochs were also excluded, predominantly as the large surface area of lochs relative to that of rivers was found to exert a disproportionately large influence on species overlap figures. Further, as beavers would likely use the perimeter of lochs and would not be capable of creating dams across lochs, beaver-salmon interactions would likely be limited relative to the area of lochs.

2.3 Data limitations

A number of assumptions and approximations were required in view of data limitations. SEPA catchment boundaries are held separately for the Rivers Tweed and Whiteadder and for the Tay and Earn since the Whiteadder enters the Tweed below the tidal limit as does the Earn in the Tay estuary. Areas considered were for the Tweed including the Whiteadder and the Tay excluding the Earn.

SNH potential beaver habitat data exclude the Tweed in England. Consequently a boundary was produced for the Tweed within Scotland and figures for the Tweed relate to the Scottish part only.

OS MasterMap data for fresh water are held as both polygons and lines, depending on river width and scale of data capture. Line data occur mainly in upper river sections, usually where stream width is less than 5m, but may also denote field ditches in lower catchment areas. Consequently, the data for some of the upper tributaries are in line format only, with no width or area information. An accurate area can therefore be calculated for river and loch sections held as polygons but not for those river sections available only as lines. In the absence of width information for upper tributaries and following investigation of suitable buffer widths, an area was created for line sections by producing 4m total width polygon buffers on the lines.

OS Mastermap freshwater polygon data contains no separation between lochs and rivers or between larger rivers and any smaller tributaries. Where salmon distribution included lochs, breakpoints were required to separate loch polygons from the rivers flowing in or out of them. These were determined using a combination of SEPA loch coding in the CEH DRN, Ordnance Survey data (section 2.1) and best judgement.

SEPA river width data were used to assist in determining the breakpoint between minor tributaries and major tributaries or main-stem categories. All streams where width values were consistently recorded as being less than 10m were categorised as minor tributaries. In certain areas, width data values were limited, and as such breakpoints were determined using best judgement and OS 1:10,000 raster data.