Figure 3: CART based regression tree of environmental variables.
From: Control factors and scale analysis of annual river water, sediments and carbon transport in China

Splits are determined by their contribution to the overall model and reduce predictive error. The longer the branches in the tree, the greater the deviance explained. (a) Rc as response variable; (b) TSSC as response variable; (c) TSSL as response variable; (d) TOCL as response variable. Abbreviations of the variables as shows in Table 1.