Indirectly accounting for uneven sampling effort to estimate population sizes from atlas data

Titeux N., Derouaux A., Jacob J.P., Kinet T., Paquet J.Y.

34 49
Read Article     Download


One important by-product of bird atlases is to provide population estimates to evaluate species conservation status. Unfortunately, large-scale atlas data are nearly always characterized by an uneven sampling effort among sampling units. This prevents from calculating straightforwardly reliable population estimates. We propose a procedure to deal with this unevenness when generating regional population estimates. The Breeding Bird Atlas of Wallonia 2001-2007 (Southern Belgium) was used to illustrate our method. Data were collected according to a grid composed of 508 sampling units. Each unit was assigned to one or several observers who gathered information on species abundance and breeding status according to an unrestricted procedure. In addition, a series of sub-sampling units were surveyed within each sampling unit to collect information according to a standardized procedure. Based on the data collected according to the unrestricted surveys in the sampling units and the corresponding standardized surveys in the sub-sampling units, we calculated three indices representing different aspects of the sampling effort at the sampling unit level. Those indices are not sensitive to the natural disparity in the nature/diversity of the habitats among the sampling units. The indices were then combined into a single score that was used to weight the contribution of each sampling unit to the calculation of the regional population estimates. Depending on the species, the consideration of the unevenness in the sampling effort led to considerable discrepancies in the population estimates.