Application of classification trees and support vector machines to model the presence of macroinvertebrates in rivers in Vietnam

a r t i c l e i n f o In the present study, classification trees (CTs) and support vector machines (SVMs) were used to study habitat suitability for 30 macroinvertebrate taxa in the Du river in Northern Vietnam. The presence/absence of the 30 most common macroinvertebrate taxa was modelled based on 21 physical-chemical and structural variables. The predictive performance of the CT and SVM models was assessed based on the percentage of Correctly Classified Instances (CCI) and Cohen's kappa statistics. The results of the present study demonstrated that SVMs performed better than CTs. Attribute weighing in SVMs could replace the application of genetic algorithms for input variable selection. By weighing attributes, SVMs provided quantitative correlations between environmental variables and the occurrence of macroinvertebrates and thus allowed better ecological interpretation. SVMs thus proved to have a high potential when applied for decision-making in the context of river restoration and conservation management. Modelling running waters based on ecological knowledge and monitoring data has proven to considerably facilitate and improve the assessment of habitats and the determination of the relationship between environmental variables and the occurrence of certain organisms. The availability of reliable data and suitable modelling techniques, which are able to handle the nonlinear and complex nature of ecosystems, resulted in the development of models with a high reliability (Recknagel, 2006). Different modelling techniques have been applied to running waters, mainly focussing on the environmental responses of river communities to specific disturbances. Modelling the distribution of taxa as a function of the abiotic environment, often called habitat suitability modelling, has been recognised as a significant component of conservation planning (Austin, 1998, 2002; Guisan and Zimmermann, 2000). Artificial intelligence has played a crucial role in studying these relationships. More recently, a range of modelling techniques has been applied for the assessment of running waters based on the distribution of macroinvertebrates. Bayesian belief networks (Adriaenssens et al., 2004b) have proven to have a high potential in macroinvertebrate habitat suitability analysis, as they combine reliable classifications with transparency. Habitat suitability models aim to relate the presence or abundance of a species at a site to environmental variables that describe their general habitat. River management can benefit from such predictive models as decision support tools to improve the efficiency of monitoring and assessment, for example by choosing the most optimal restoration measure from a set of given river restoration scenarios by predicting the effect …