Skip to main content
. 2018 Aug 9;13(8):e0201558. doi: 10.1371/journal.pone.0201558

Table 2. Table of the forward model selection results with the base phylogeny model and the top models from the model selection process, based on minimum ΔAIC.

Model ΔAICa Log-Likelihood K
Abundance+Timing+Week+(1|Family/Genus/Species) 0 -633.77 9
Abundance+Timing+(1|Family/Genus/Species) 0.6 -635.03 8
Abundance+Timing+Week+Migrant+(1|Family/Genus/Species) 3.0 -633.26 11
Abundance+Timing+Week+Forage+(1|Family/Genus/Species) 3.2 -629.34 15
Abundance+Timing+Migrant+(1|Family/Genus/Species) 3.3 -634.42 10
Abundance+Timing+Forage+(1|Family/Genus/Species) 3.4 -630.45 14
Abundance+(1|Family/Genus/Species) 16.2 -644.83 6

Model descriptions show the variables used in each model to predict Collisions using a negative binomial regression function. Of the 9 models calculated in the analysis, only the models which had ΔAIC<9.0, are shown, in addition to the starting model of Abundance + (1|Family/Genus/Species). The model with the lowest AIC was Abundance+Timing+Week+(1|Family/Genus/Species). The model Abundance+Timing+(1|Family/Genus/Species) was highly competitive. K represents the number of parameters in each model.

aΔAIC calculated based on minimum AIC value amongst all models. Minimum AIC value was 1285.5 for the model Abundance+Timing+Week+(1|Family/Genus/Species).