Home > Research > Services > Report Abstracts > 217

Abstract from BTO Research Report No 217:

Field, R.H. & Gregory, R.D. (1999)

Measuring population changes from the Breeding Bird Survey.

Executive summary

1. The BTO/JNCC/RSPB Breeding Bird Survey (BBS) involves volunteers making bird counts in randomly selected Ordnance Survey (OS) 1-km squares using a line transect method. Within each 1-km square volunteers are asked to establish two parallel line transects oriented north-south or east-west. The line transects should be 500m apart and 250m from the edge of the square. The 2-km of transect are divided into ten 200m sections. Three visits are made to each transect in a season, the first to record the habitat, and then early and late season visits to record the number and species of birds present. This report uses data collected between 1994 and 1997.

2. BBS primarily monitors temporal changes in British breeding bird populations by calculation of annual population indices. It is essential that these indices reflect the real populations from which the data are collected. Here we assess the statistical models used to calculate these indices and the influence of environmental variables, timing of sampling visits and occurrence of flocks and passage of birds upon them.

3. The timing of early and late visits to BBS squares was very consistent between years, with a slight tendency to a wider spread of dates in 1997. Early counts tended to sample residents and early migrants whilst late counts sampled summer migrants. Therefore, the plasticity of visit timing within local BBS schemes appears to be sufficient to allow good sampling of common breeders. These results justify the use of maximum counts for the calculation of indices.

4. Correlations of species counts with weather conditions were highly variable and often appeared contradictory. Recalculation of indices after removal of counts made under non-ideal, but acceptable, weather conditions was felt to be unreliable. BBS will probably benefit more from the rigid adherence to not counting in adverse weather conditions. The influence of weather appears to vary between species and more detailed analyses incorporating year and site effects would be beneficial once more data are available.

5. Removal of flocks of greater than ten birds per 200m section produced reliable revised indices for the seven commonest wader species. That for Golden Plover also incorporated the removal of birds on non-upland (by ITE land classification) squares. Wildfowl indices were recalculated after flock removal but these generally showed little improvement in standard errors and are probably no more reliable than the standard indices. However, some screening for very large counts would seem appropriate. Gull indices were not revised due to the aggregated nature of breeding in these species and the presence of groups of non-breeding birds. The BBS is unlikely to be the best source of information on population changes for the latter two groups.

6. The standard BBS index model (site x year) was assessed in relation to a linear trend (site x year count) index model. Both indicated broadly similar population changes over the first four years of BBS data. The linear trend model may become more reliable when more years' data are included in index calculations. The use of different models largely depends upon what is being assessed. If it is the overall trend that is of interest, a linear trend, or more appropriately a non-linear trend, model should be adopted, although we need many more years of data before this would be useful. BBS coverage of the commoner species is adequate to produce reliable estimates of population changes, but coverage should be increased further, or targeted towards less frequent species to enhance their monitoring.

Back to Research Reports 159-247