Exploring the effects of truncated,pooled and sexed data in distance sampling estimation of breeding bird abundance

Capsule Field methods used by the UK's Breeding Bird Survey (BBS) provide a practical approach to estimating breeding bird abundance and this paper discusses how they might be adapted to increase accuracy.

Aims Using Salisbury Plain as a case study, examine the use of distance sampling to produce estimates of breeding bird abundance.

Methods During 2005, 157 1 km squares were surveyed on Salisbury Plain using the UK’s BBS methods, with the exception that all birds were sexed whenever possible and allocated to one of five distance bands (0–10 m/10–25 m/25–100 m/100–250 m/ >250 m). Data were modelled to investigate the effects of truncated, pooled and sexed bird data in estimating breeding populations.

Results Pooling of the inner distance bands made little difference to estimates, while truncation over 100 m affected them by up to 25%, generally leading to an increase in density and the width of the confidence limits. The national BBS does not distinguish between bird sexes and therefore any density estimate produced relates to individuals. Our analysis suggests that halving the number of individuals (assuming an equal sex ratio) could lead to significant underestimation of population. This is particularly the case for species where there are differences in detectability between sexes or skewed sex ratios. In such cases, the density of males may represent a more accurate assessment of the population.

Conclusions Bird surveys incorporate distance sampling because it assesses changes in detection probability and this paper suggests how these methods might be adapted to increase accuracy. Changes include increasing both survey effort (e.g. number of visits or distance covered), the number of distance bands, and recording sexed bird data. Numbers of pairs or territories could then be estimated by either halving the density of individuals, or using the density of males, whichever is highest. These suggestions have not been tested against a known population and further work is desirable.