Moose–tree interactions: rebrowsing is common across tree species

BMC Ecology

Table 2 Overview of variables included in linear and generalised linear mixed models to test each prediction

Prediction	Response variable	Predictor variables	Random intercept
i.1	Probability of browsing (0/1)	ABI, moose, prod	Area/stand/plot
i.2	Ln(browsed.shoots)	ABI, moose, prod, ln(av.shoots + 1)	Area/stand/plot
i.3	Bite diameter (mm)	ABI, moose, prod	Area/stand/plot
ii.1	Probability of browsing (0/1)	ABI*sp, moose, prod	Area/stand/plot
ii.2	Ln(browsed.shoots)	ABI*sp, moose, prod, ln(av.shoots + 1)	Area/stand/plot
ii.3	Bite diameter (mm)	ABI*sp, moose, prod	Area/stand/plot
iii.4	Ln(av.shoots +1)	ABI*ln(tree height), prod	Area/stand/plot
iii.5	Tree height (standardized)	ABI*stand height, prod	Area/stand/plot
iii.6	Shoot diameter	ABI*sp, height above ground	Plot/tree ID
iii.7	Shoot length	ABI*sp, height above ground	Plot/tree ID

Predictions i.1–ii.3 investigate the moose response (current browsing) to accumulated browsing (ABI), while predictions iii.4–7 investigate the tree’s morphological response to previous browsing. Prediction i.1–3 were analysed separately for each individual tree species. Prediction ii.1–3 and iii.4–5 were analysed for birch and pine only, because they provided sufficient data. For prediction iii.6–7 all tree species were grouped together, excluding spruce and alder due to insufficient data
Sp species, moose moose pellet groups, prod productivity index from vegetation type, av. shoots available shoots in browsing height (0.5–3 m)

ISSN: 1472-6785