Skip to main content

Table 3 Model performance

From: Variables related to nitrogen deposition improve defoliation models for European forests

Model

No. of plots

No. of LVs

Xvar (%)

Yvar (%)

RMSEC

Beech

Reference

20

1

25.3

4.6

1.28

Reference + BADoAP

20

3

66.8

39.0

1.07

Reference + Nthrdep

20

1

14.3

19.5

1.21

Reference + BADoAP + Nthrdep

20

2

28.4

49.3

0.96

Referencea

18

1

36.2

19.8

1.13

Reference + foliar

18

2

38.1

60.8

0.83

Referencea

16

1

39.5

6.2

1.24

Reference + soil

16

1

19.5

47.9

0.96

Norway spruce

Reference

33

1

46.0

8.1

1.03

Reference + BADoAP

33

1

36.6

16.7

1.03

Reference + Nthrdep

33

1

42.6

12.0

1.02

Reference + BADoAP + Nthrdep

33

1

35.6

19.5

1.00

Referencea

29

1

48.5

6.3

1.04

Reference + foliar

29

1

36.3

37.2

0.93

Referencea

24

1

53.9

8.4

1.07

Reference + soil

24

1

35.4

17.0

1.15

Scots pine

Reference

18

2

66.1

29.0

1.10

Reference + BADoAP

18

2

63.4

25.2

1.07

Reference + Nthrdep

18

5

92.0

69.3

0.92

Reference + BADoAP + Nthrdep

18

5

92.0

69.6

0.96

Reference + foliar

18

1

29.4

35.4

1.03

Referencea

16

1

46.8

44.5

0.84

Reference + soil

16

4

80.5

76.9

0.74

  1. Number of plots, number of latent variables (LVs), variance explained on the set of predictors (Xvar, %), variance explained on the response (Yvar, %), root mean square error in cross-validation (RMSEC)
  2. BADoAP biotic or abiotic damage from known causes other than air pollution, Nthr dep nitrogen throughfall deposition
  3. aModels built on a reduced data set to compare with the model presented in the following row of this table
Back to article page

Annals of Forest Science

ISSN: 1297-966X