Modelling the disappearance of coarse woody debris, following a land clearing event

Table 5 Some of the combinations of explanatory variables tested for the generalised additive mixed-effects model in Eq. (1), and the performance when applied to withheld validation data

Model	Combination of explanatory variables	PV	CCC	PCP	MAE	\({\varvec{L}}^{2}\)
1	\(c + b + {\text{s}}\left( {t\|c} \right) + {\text{s}}\left( {r,v} \right) + g + {\text{s}}\left( a \right)\)	0.33	0.47	0.80	13.7	17.6
2	\(c + b + {\text{s}}\left( {t\|c} \right) + {\text{s}}\left( {r,v} \right) + g\)	0.33	0.48	0.81	15.6	14.7
7	\({\varvec{c}} + {\varvec{b}} + {\mathbf{s}}\left( {{\varvec{t}}\|{\varvec{c}}} \right) + {\varvec{g}}\)	0.32	0.46	0.82	14.7	10.5
12	\(c + b + {\text{s}}\left( {t\|c} \right)\)	0.32	0.45	0.81	14.0	14.2
13	\(c + {\text{s}}\left( {t\|c} \right) + {\text{s}}\left( {r,v} \right)\)	0.31	0.45	0.83	14.9	11.5
24	\(b + {\text{s}}\left( t \right) + {\text{s}}\left( {r,v} \right)\)	0.30	0.44	0.82	16.6	18.8
36	\({\text{s}}\left( t \right)\)	0.25	0.36	0.83	16.0	27.0
37	(none)	0.00	0.00	0.91	25.8	33.9

Function ‘\({\text{s}}\left( . \right)\)’ indicates a penalised cubic regression spline. Key to the explanatory variables: c = clearing method; b = bioregion; t = decimal years since clearing; r = proportion of rain days since clearing; v = mean daily vapour pressure deficit since clearing; g = number of times burned since clearing; a = clay content of soil surface (0–5 cm). Refer to Tables 2, 3, 4 for more information on the explanatory variables. Key to the columns: PV proportion of variance explained (logit scale), CCC concordance correlation coefficient (logit scale), PCP proportion of sites correctly predicted, MAE median absolute error (original scale of %); \(L^{2}\) Euclidean norm. Bold-face indicates the best overall model

ISSN: 1750-0680