Countryside Survey 2000 Module 4: Increasing upland representation in Countryside Survey 2000 - FINAL REPORT

Countryside Survey 2000 has generated estimates of the extent of Broad Habitats in the UK and at country level. The sampling units used in Countryside Survey 2000 (CS2000) are 1 km squares and the sample is stratified according to the ITE Land Classification. For reasons associated with policy requirements, particularly in Scotland, the land classification was developed for CS2000 so that there were separate classes in Scotland and in England with Wales. A result of this rearrangement of the sampling strategy was that those land classes which occurred within the uplands of England and Wales (Environmental Zone 3) were not particularly well sampled (54 squares at a sampling rate of 1: 283). Both to ensure adequate representation of each of the ITE Land Classes in Environmental Zone 3 (EZ3), and to improve the sampling rate in the uplands in particular, 38 new squares were added to those squares already sampled in CS1990, to form the sample for CS2000. Of these ‘new’ squares, 20 were funded by MAFF. Although it is not possible to tease out the effects of just the extra 20 squares, this report does address two key questions relating to the 38 new squares in EZ3: 1. by how much is the statistical robustness of estimates of Broad Habitats improved buy the addition of 38 new squares? 2. is the new sample of squares different from the old sample, in terms of Broad Habitat composition?, Statistical robustness The area of Broad Habitat in Environmental Zone in 1998 (the survey year for CS2000) was estimated using (a) CS1990 sample squares only (b) CS1990 squares plus the ‘new’ squares. The statistical error rates, as represented by the Coefficients of Variation of the estimate, were all reduced, most by at least 20%. Six of the Broad Habitats, representing about 64% of the area of EZ3, had CVs of less than 20% (as opposed to just two when the CS1990 sample was used alone). Representativeness of the ‘new’ squares A number of statistical tests were performed to compare the Broad Habitat characteristics of different samples of squares. The most appropriate test (Kolmogorov-Smirnov Test comparing the 54 old squares with the 38 new squares) showed no statistically significant differences between the two populations of squares. The least similar category, as judged from other tests, was Bog. In summary, the 38 additional sample squares have improved the statistical robustness of the estimates of Broad Habitats in Environmental Zone 3 (by at least 20%) and, in terms of average Broad Habitat composition, have been shown to be no different from other squares in the zone.