Ecological Archives E093-248-A4
Jessy Loranger, Sebastian T. Meyer, Bill Shipley, Jens Kattge, Hannah Loranger, Christiane Roscher, Wolfgang W. Weisser. 2012. Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in experimental monocultures. Ecology 93:2674–2682. http://dx.doi.org/10.1890/12-0328.1
Appendix D. Graph of the results from the random forest (RF) analysis, explaining how the selection of traits was done.
Fig. D1. Graph of the results from the Random Forest (RF) analyses to select which plant traits are of importance to predict the invertebrate herbivore damage in the monocultures (HDM) of the Jena Experiment (Germany). Each point is a trait and the x-axis represents the order of the traits according to their importance score value which is represented by the y-axis. The higher a trait's score is the more important is the trait to predict HDM. All traits (13) to the left of the line were selected as being of importance; all traits to the right were discarded. The selection was based on the curve illustrated here, looking at the decrease in importance between traits. The last relatively large drop in importance was after the 12th trait, thus the first 12 traits should have been selected. However, the five following traits were also tested with a multiple regression to confirm that traits could be excluded from this point. Only the 13th trait proved to be of importance and thus was also kept for the next step of the analyses.
[Back to E093-248]