Published June 17, 2019 | Version v1
Journal article Open

Phenolic compounds as unambiguous chemical markers for the identification of keystone plant species in the Bale Mountains, Ethiopia

Authors/Creators

  • 1. Martin Luther University Halle–Wittenberg
  • 2. Technical University of Dresden
  • 3. University of Bayreuth
  • 4. Addis Ababa University

Description

Figure S1: Comparison of Accuracy (A) and F1 Score (B) of Support Vector Machine (SVM, blue), Random Forest (RF, red) and Recursive Partitioning (RP, grey) algorithm based on the level of relative phenols, in relation to the ratio of tested/total number of samples (n=47) that has been split into test and training datasets, each model has been computed 5 times, the bars indicate the standard derivation between the prediction results of the models; Figure S2: Principal Component Analysis, PCA based on the relative phenol abundance in 47 leaf and twig samples from the Bale Mountains, shown are the first two principal components (PC1 and PC2). Dark red arrows indicate the direction of each vector of feature; Figure S3: Cross-validation of the model used; Table S1: Sum of weighted mean of phenolic compounds of each dominant plant species.  

Files

Figure S1A (Accuracy).jpeg

Files (473.4 kB)

Name Size Download all
md5:3787ab238983dab2813142cc50c6ec7a
137.9 kB Preview Download
md5:0cbebda0d39d95066d1c0ca2fba3375b
139.3 kB Preview Download
md5:fa0ab7537595014498987db8ed7f2a54
113.9 kB Preview Download
md5:d6fc012e2316dae99c7464f620968705
30.3 kB Preview Download
md5:7e7a3cce0244e6b545935a264d29d777
51.9 kB Download