Home > Research > Services > Report Abstracts > 303

Abstract from BTO Research Report No 303:

Freeman, S.N., Noble, D.G., Newson, S.E. & Baillie, S.R. (Oct 2002) Published 2003.

Modelling bird population changes using data from the Common Birds Census and the Breeding Bird Survey.

1. Population trends of common and widespread terrestrial birds in the UK are derived from the two main sources of data; the Common Birds Census (CBC) run from the early 1960s to 2000, and the BTO/JNCC/RSPB Breeding Bird Survey (BBS) run from 1994 to the present. In this report, we compare species trends from both surveys during the period of overlap, and in the different geographical areas covered by these surveys, in order to assess the potential for producing long-term combined CBC-BBS trends.

2. Based on a comparison of CBC and BBS trends for 73 species that are routinely monitored by both surveys, it is demonstrated that combined CBC/BBS indices can be produced for the period of overlap (1994 to 2000) for 66 of these species within Southern Britain (defined as a square bounded by easting 3000 and northing 5000), where the majority of CBC survey effort is focused. For these species, the difference in trends between the two surveys is not significant. The remaining seven species show significantly different CBC and BBS trends within Southern Britain. For four of these species, differences are likely to be related to a high level of power resulting from the high species abundance in the case of the Wren, Robin, Blackbird and Chaffinch. However, a real problem may exist in combining CBC and BBS trends for the Stock Dove, Pheasant and Chiffchaff and possibly also Coal Tit.

3. A comparison of BBS trends within and outside Southern Britain (defined as a square bounded by easting 3000 and northing 5000) shows that for a significant proportion of species (38 of 73 species: 52%), trends within Southern Britain are not representative of trends elsewhere. Therefore CBC data should not automatically be used to produce population indices for the UK as a whole. In this study, Southern Britain was chosen as an area within which to compare trends, based on work by Fuller et al. 1985, which showed that farmland CBC plots at least were representative of this land area. It could be argued that calculating trends within England, which covers a broadly similar area, may be more appropriate. In this case, boundaries should be agreed before routine analyses is introduced and trends examined in comparison with Southern Britain as defined here.

4. Results show that it is possible to produce combined CBC-BBS trends for Southern Britain for the majority of species monitored. A comparison of trends within and outside the area of good coverage of the CBC suggests that for about half of the species tested, a combined CBC-BBS index using data from the entire UK could be justified, on the assumption that regional differences in trends during the 1990s were similar to those in the preceding decades. When producing combined CBC/BBS indices, data from both surveys for the full overlap period should be used. The reason for this is that if a single or small number of years were used for ‘attaching’ trends from the two surveys and that these years were not representative of the underlying trend, parts of the resulting index are likely to be biased.

5. A number of points were raised as a result of this work with respect to future combining of CBC and BBS indices. Further work would be needed to evaluate the use of weighting within any combined indices for which a simulation using hypothetical data may be an appropriate approach. In terms of optimising precision and smoothing, the use of Generalised Additive Models (GAMs) should be explored. However, the main problem is the large amount of computing time that would be required to calculate, via the bootstrap method, standard errors in such analyses.

Back to Research Reports 289-344