Table 1 Relative importance for prediction (%) of the macroclimatic variables, goodness-of-fit value (AUC) and the goodness-of-fit value ('AUC with coordinates') if macroclimate is replaced by coordinate-related variables for all the five 'region' models. Variables with high contribution (> 20%) are bolded.

From: Empirical delineation of the forest-steppe zone is supported by macroclimate

Region

Aridity

Annual mean temperature

Isothermality

Temperature seasonality

Annual precipitation

Precipitation of driest quarter

Precipitation of warmest quarter

AUC

AUC with coordinates

Southeast Europe

2.77

38.20

–

22.27

3.63

27.34

5.79

0.979

0.726

East Europe

21.55

7.92

14.84

10.15

–

41.06

4.49

0.953

0.981

West Siberia

26.71

28.95

3.75

5.55

–

32.14

2.89

0.953

0.913

Inner Asia

13.60

30.22

10.85

13.99

–

11.75

19.60

0.922

0.929

Far East

18.07

22.13

–

13.04

17.22

15.35

14.19

0.981

0.953

  1. Further details (i.e., learning rate and number of trees) can be found in Table S2.1.
Back to article page