• English 
  • Spanish 

Predicting plant species distribution across an alpine rangeland in northern Spain. A comparison of probabilistic methods

Journal: Applied Vegetation Science
Year: 2011  
Status: Published
In this status since: 28 Jan 2011
PDF file: Bedia_et_al_2011_AVSCI.pdf
DOI: 10.1111/j.1654-109X.2011.01128.x
, Busque, J.,

Predictive models constitute an important tool in ecology. Using presence/absence data of 15 plant species of an alpine rangeland in northern Spain, and a set of 14 topographical and geomorphological descriptors of relatively easy acquisition, we examined and compared the performance of five state-of-the-art methods used in ecological modeling: Multiple Logistic Regression (MLR), Artificial Neural Networks (ANN), Support Vector Machines (SVM), Classification and Regression Trees (CART), Maximum Entropy (MAXENT) and Multivariate Adaptive Regression Splines (MARS). Validation of the models was carried-out computing the Area Under the ROC Curve (AUC) using leave-one-out cross validation and the resolution and reliability diagrams of the resulting probabilistic predictions. We also analyzed the binary presence/absence deterministic predictions obtained by setting two different probability thresholds: the species prevalence and a ROC-optimized value, and we computed the corresponding confusion matrices to calculate sensitivity, specificity, Cohen’s kappa and the True Skill Statistic (TSS). The overall result of this comparison shows that the performance of each technique varies depending on the target species; in general, CART exhibited a poor performance and MLR was competitive with the more sophisticated ANN, MARS and SVM methods. The best predictive resolution was obtained in most cases by ANN followed by SVM and CART models; on the other hand, MLR and MARS were generally the best calibrated. We also present an ecological interpretation of results, with emphasis in the possible ways of improving our models. Most of the target species were accurately predicted evidencing that geomorphological and topographical variables are suitable descriptors at the scale of analysis.