Home > Research > Services > Report Abstracts > 318

Abstract from BTO Research Report 318:

Newson, S.E. (2003)

Producing statistically valid maps of species abundance using geostatistical analysis in ArcGIS.

Executive summary

We examine the potential of the Geostatistical Analyst extension of ArcGIS for interpolating statistically valid maps of species abundance from survey data. To explore this methodology, we use Breeding Bird Survey (BBS) data for 2000, covering 11 species ranging from widespread and abundant to rare and localised species.

The results demonstrate that it was possible to produce maps that matched well the expected distribution and abundance for the majority of species. However it was not possible to produce maps for Willow Tit and Nightingale, which are poorly monitored by the BBS because they occur at low densities and are highly localised in their distribution. Further to this, predictions of abundance for species that have specific habitat requirements and show a restricted range, such as Reed Warbler and Nuthatch based purely on location, are likely to be improved by narrowing the area over which predictions are made, and may benefit from co-kriging models which include habitat as a predictor variable. Alternatively presence/absence could be modeled using indicator kriging.

Examining the potential of this methodology for producing automated production of maps it was encouraging to find that models with default parameters chosen by the program compared well with predictions from manual diagnoses of the data and modelling. However, there is some reduction in the level of precision that will reduce the number of species for which abundance maps can be produced.

In addition to co-kriging and indicator kriging mentioned above, further work could use this methodology to model the temporal as well as spatial change in species abundance or distribution, providing a means of visually identifying geographic areas of significant population change, perhaps prior to further data analysis.

Back to Research Reports 289-344