Fig. 2

Classification (left) and regression (right) trees from analyses to assessing the influence of S. scabiei infestation, environment and host sex on the prevalence and abundance of lice and ticks. Each node represented by the variable abbreviation (variable names as shown in Table 2) defines a split of the dataset and the branches define the path taken by each observation. Ellipses represent the terminal nodes of the tree model. The terminal nodes of classification tree models (left) indicate if an ectoparasite species is present (1) or absent (0) in relation to S. scabiei infestation, environment and host sex. The terminal nodes of regression tree models (right) express the variations in ectoparasite abundance in relation to the aforementioned variables. The prediction error rate for each model is presented in bold. Abbreviations: PERCT, prediction error rate for classification trees; PERRT, prediction error rate for regression trees