Skip to main content

Table 2 Candidate models and akaike weights

From: Patch-use dynamics by a large herbivore

  Candidate models AIC DeltaAIC AICw
Model M* ~ TmKnown*NDVI + Ruggedness + Herd*(DistRd), rand(ElkID) 2857.8 0.0 0.77
Model L* ~ TmKnown + Ruggedness + Herd*(DistRd) + NDVI, rand(ElkID) 2860.7 2.9 0.18
Model K* ~ TmKnown + Ruggedness + Herd*(DistRd), rand(ElkID) 2863.1 5.3 0.05
Model G* ~ TmKnown + NDVI + Herd*(DistRd + Traffic), rand(ElkID) 2878.4 20.7 0.00
Model J ~ TmKnown + Ruggedness, rand(ElkID) 2882.4 24.6 0.00
Model H* ~ TmKnown*NDVI, rand(ElkID) 2884.4 26.7 0.00
Model I* ~ TmKnown*Aspect, rand(ElkID) 2887.7 29.9 0.00
Model E ~ TmKnown + NDVI, rand(ElkID) 2888.3 30.6 0.00
Model F ~ TmKnown + NDVI + Aspect + Canopy, rand(ElkID) 2891.1 33.3 0.00
Model D* ~ NDVI + Herd*(DistRD + Traffic), rand(ElkID) 3180.7 323.0 0.00
Model C ~ NDVI + Aspect + Canopy, rand(ElkID) 3187.6 329.8 0.00
Model A ~ NDVI, rand(ElkID) 3193.0 335.2 0.00
Model B ~ NDVI + Aspect, rand(ElkID) 3195.0 337.2 0.00
  1. *All models with interaction effects included main effect terms of interacting covariates.
  2. The candidate model set contained 13 models comparing the influence of vegetation, physiogeographic and disturbance variables. The top model included ruggedness, TmKnown, distance to road, and productivity, receiving 77% of support in the data.