Ecological Archives A025-143-A3

Iain M. McNicol, Casey M. Ryan, and Mathew Williams. 2015. How resilient are African woodlands to disturbance from shifting cultivation? Ecological Applications 25:2320–2336. http://dx.doi.org/10.1890/14-2165.1

Appendix C. Summary and comparisons of all the statistical models fit to the data.

Table C1. Goodness of fit statistics for each of the statistcal models fit to each data set. Model selection was based on ANOVA tests where models were nested (linear vs. quadratic), or the minimisation of the Akaike Information Critetion (AICc), corrected for small sample sizes, and the residual standard error (RSE) when comparing models fit using linear (‘lm’ function in R) and non-linear (‘nls’) methods. R-sqaured values are provided where it is an acceptable goodness of fit measure. Model forms followed by the letters ‘n.s’ are those which had insignificant parameter estimates and so were excluded from consideration.

Ecosystem component

Soil type

Remant trees

Fitting method       (R command)

Model form

n

df

ANOVA       (F, p value)

AICc

Residual sum of squares (*104 )

RSE

r²

Aboveground carbon stocks

Sand

Yes

OLS (lm)

Linear

31

29

0.19; 0.74

238

n/a

10.4

0.5

Yes

OLS (lm)

Polynomial

31

28

240.53

n/a

10.57

0.48

Yes

Robust (rlm)

Linear

31

29

239.2

n/a

10.31

Clay

Yes

OLS (lm)

Linear

21

19

2.64; 0.12

151.99

n/a

7.95

0.66

Yes

OLS (lm)

Polynomial

21

18

152.2

n/a

7.63

0.68

Yes

Robust (rlm)

Linear

21

15

152.019

n/a

5.836

Sand

No

OLS (lm)

Linear

31

29

185.19

n/a

4.44

0.84

Clay

No

OLS (lm)

Linear

21

19

129.95

n/a

4.706

0.74

Soil carbon stocks

->  Topsoil (0 - 5cm)

Sand

OLS (lm)

Linear (n.s)

31

29

144.64

n/a

2.308

0

Clay

OLS (lm)

Linear (n.s)

20

18

109.48

n/a

3.266

0.13

NLR (nls)

Negative exponential (n.s)

20

18

109.73

n/a

3.286

->  Subsoil (5 - 30cm)

Sand

OLS (lm)

Linear (n.s)

31

29

224.85

n/a

8.415

0

Clay

OLS (lm)

Linear (n.s)

20

18

190.39

n/a

24.68

0.01

NLR (nls)

Negative exponential (n.s)

20

18

190.39

n/a

24.68

Tree species diversity

->  Species richness

Sand

Yes

OLS (lm)

Polynomial

24

21

170.12

28.52

7.25

0.46

Yes

NLR (nls)

Saturation

24

22

168.57

29.34

7.29

Yes

NLR (nls)

Unimodal (ricker)

24

22

166.79

28.32

7.02

Clay

Yes

OLS (lm)

Polynomial

17

14

137.68

12.74

10.96

0.30

Yes

NLR (nls)

Saturation

17

15

138.93

13.16

12.17

Yes

NLR (nls)

Unimodal (ricker)

17

15

134.49

13.66

10.68

-> Fisher's alpha

Sand

Yes

OLS (lm)

Polynomial

24

21

128.69

3.53

3.06

0.16

Yes

NLR (nls)

Saturation

24

22

125.53

3.53

2.97

Yes

NLR (nls)

Unimodal (ricker)

24

22

128.06

3.09

3.13

Clay

Yes

OLS (lm)

Polynomial

17

14

107.39

1.66

4.50

0.23

Yes

NLR (nls)

Saturation

17

15

108.20

1.69

4.93

Yes

NLR (nls)

Unimodal (ricker)

17

15

104.66

1.74

4.44

Tree species composition

Sand

Yes

OLS (lm)

Linear

62

60

1.01; 0.316

-109.02

108.37

0.096

0.29

Yes

OLS (lm)

Polynomial

62

59

-107.79

108.37

0.096

0.29

Yes

NLR (nls)

Logarithmic

62

60

-111.48

110.39

0.095

Clay

Yes

OLS (lm)

Linear

26

24

0.49; 0.486

-22.71

24.485

0.142

0.16

Yes

OLS (lm)

Polynomial

26

23

-20.46

24.485

0.143

0.15

Yes

NLR (nls)

Logarithmic (n.s)

26

24

-21.26

22.81

0.146

Sand

No

OLS (lm)

Linear

56

54

0.13; 0.71

-94.73

100.87

0.099

0.21

No

OLS (lm)

Polynomial

56

53

-92.54

100.87

0.1

0.19

No

NLR (nls)

Logarithmic

56

54

-95.29

101.26

0.093

Clay

No

OLS (lm)

Linear

26

24

0.24; 0.62

-23.35

23.33

0.14

0.19

No

OLS (lm)

Polynomial

26

23

-20.81

23.33

0.142

0.16

 

 

No

NLR (nls)

Logarithmic (n.s)

26

24

-22.29

21.76

0.143


[Back to A025-143]