The Breeding Bird Survey for mapping Britain's birds: A preliminary assessment of performance at two spatial scales

Advances in the application of geostatistics in recent years have improved the precision of predicting occurrence or relative abundance at non-surveyed sites and so allow the potential for producing reliable maps over the entire area of interest. We evaluate the use of geostatistics for producing statistically valid maps of relative abundance at two spatial scales from annual bird monitoring data collected by the UK Breeding Bird Survey (BBS). Whilst development of an approach that optimizes the precision of this methodology is being explored currently, it is important to explore the extent to which the BBS, with about 2,000 1-km squares surveyed annually (<1 % of the UK), is adequate for producing useful maps. To this end, we compare relative abundance, interpolated from 2003 BBS data for 96 species, with independent data collected for these species through intensive survey effort during the last breeding Atlas of 1988-91. Comparisons are made at two spatial scales: at a regional scale with 125 regions and at a local scale comprising some 2,882 10-km squares. Whilst the BBS and Atlas measures of abundance are not the same, and there are over ten years between the collection these data, during which time several species have undergone large population declines or increases and contractions or expansions in range, it was encouraging to find that there was no significant difference at a regional scale between measures of abundance for 84 of the 96 species (88%). At a local scale, there was no significant difference between measures of abundance for 17 of the 96 species (18%). Differences between the interpolated BBS and Atlas measures of abundance at both spatial scales can be explained in part by real changes in abundance between years although, as one would expect, the BBS provides greatest map reliability for the most widespread species as shown by a strong positive relationship between map similarity and number of occupied BBS squares. However, we have yet to examine the extent to which habitat and other landscape variables may improve our predictions for species with patchy distributions.