Published online by Cambridge University Press: 14 November 2012
Effective conservation planning must anticipate the rates and patterns of dynamic threats to biodiversity, such as rapid changes in land use. Poor understanding and prediction of drivers and patterns of conversion of habitat can hinder assessments of the relative vulnerability of areas of remaining indigenous habitat to conversion, and identification of habitats in most immediate need of protection. Methods developed to model vulnerability to conversion vary in their complexity and applicability to conservation management. Generalized additive models provide a simple robust method to explore predictors and patterns of land-use conversion, and may be used to predict future patterns of conversion using recent land conversion data. This paper provides the first data-derived and statistically validated measurement of the vulnerability of New Zealand's indigenous grasslands to conversion. Higher altitude and more marginal (for agriculture and forestry) land showed greater conversion, and models based on earlier conversion patterns performed more poorly in predicting current patterns of conversion. Up-to-date land conversion data appear crucial for accurately predicting future vulnerability to habitat conversion.