Abstract from BTO Research Report No 356:

Newson, S. (2006)

Comparison of London wild bird population trends with those in the surrounding area: update 1994-2002
ISBN: 1-904870-84-8

Executive summary

1. INTRODUCTION

In response to a previous request from Dr D.G. Dawson of the Greater London Authority, data from the BTO/JNCC/RSPB Breeding Bird Survey (BBS) for 1994-2000 were analysed in order to statistically compare bird population trends in Greater London with those in the surrounding area defined as the Southeast and East of England Government Office regions combined (Newson & Noble 2003). The results, in a series of tables and figures, present population trends with confidence intervals as well as a test of the significance of differences in the linear trend between London and the surrounding area for the period 1994 to 2000. These results were provided to the Greater London Authority for inclusion in their state of the Environment Report. This report updates these analyses to compare bird trends in Greater London with those in the surrounding area for the period 1994 to 2002.

2. ANALYTICAL METHODS

We examine trends in 21 wild bird species reliably monitored in Greater London and the surrounding area (the Southeast and East of England Government Office Regions) over the period 1994 to 2002. Foot-and-mouth had no apparent influence on survey coverage within London in 2001, whilst Southeast and East of England was severely affected, reducing and biasing coverage in this year. In these analyses, we include data for 2001 for Southeast and East of England (combined), and make the assumption that data from a single year in a run of nine will have relatively little influence on the underlying linear trend. However, caution should be made when examining the annual indices for 2001, particularly for ‘outside London’ because that index is almost certainly biased towards trends on non-farmland sites.
However, inclusion of 2001 data in the annual model should have very little effect on the estimates for any other year. The only possible influence on estimates for other years would come from the inclusion of sites covered in 2001 and in only one other year – for example sites started in 2001 and continued in 2002. The number of these in this region is tiny (<10) and even if trends were radically different on these sites, they should have little impact on the estimates.

In all cases, data were analysed using log-linear Poisson regression models fitted in SAS (SAS 1996). Annual indices we first produced by modeling a matrix of annual site counts, with site and year effects (ter Braak et al. 1994). The year effect is an annual index of total numbers, whilst the site effect describes how species abundance at sites differ from one another. The last index of a run of years is set to an arbitrary value 1 and other indices are measured relative to this.

The second analyses examines the significance of the underlying trend by removing the year effect from the model and formally testing whether the trend is significantly different from zero. We then establish whether linear trends within London and Southeast and East of England combined are significantly different from another. In these analyses, an additional variable ‘region’ was employed where ‘region’= 1 for a site in the London Government Office Region and 0 in the Southeast or East of England Government Office Regions. Adding to the model a year*region interaction enables a formal assessment of the significance of the difference in linear trends, between two geographic areas for the time period 1994 to 2002.

Back to Research Reports 345 onwards