Modelling soil series data to facilitate targeted habitat restoration: a polytomous logistic regression approach

Abstract

Habitat restoration at a landscape scale is becoming increasingly important in environmental management. In this context, geographical information systems are well suited as they can store and integrate many of the abiotic and biotic criteria used to assess the ecological worth of a site. However, this capacity can be limited by the availability or suitability of spatial data sets. A classic example of the latter case is the National Soil Map of England and Wales, which groups soils of a varied nature into associations. Consequently the national soil map has proved to be a poor predictor of habitat suitability. Using polytomous logistic regression we put forward a method for separating soil associations into their constituent soils within the Chilterns Natural Area. This approach used soil association, aspect, slope and relative height as variables for this analysis. Whilst the model's performance is likely to have been limited by the accuracy of the soil association data set, a predictive accuracy of between 47 and 65% is sufficient to facilitate better targeting of habitat restoration when combined with other abiotic factors such as climate and topography.