Table 4 Model ranking of the move persistence mixed effect models based on tracking data for male walruses tagged in Svalbard, Norway.

From: Interannual site fidelity by Svalbard walruses

Season

Model formula

df

ΔAIC

ΔLR

Summer

 ~ ice * depth + (depth | id)

7

 − 423,976.4

211,995.2

 ~ ice + depth + (1 | id)

5

74.49

39.25

 ~ depth + (depth | id)

7

184.52

92.26

 ~ depth + (1 | id)

5

253.93

126.96

 ~ ice + (ice | id)

7

1062.31

573.99

 ~ ice + (1 | id)

5

1143.98

531.15

 ~ 1 + (1 | id)

5

1164.06

585.03

Winter

 ~ ice * depth + (depth | id)

7

 − 163,578.2

81,794.83

 ~ ice + depth + (ice + depth | id)

10

8.50

1.25

 ~ ice * depth + (1 | id)

5

76.47

40.23

 ~ ice + depth + (ice | id)

7

77.56

38.78

 ~ depth + (depth | id)

7

123.15

61.57

 ~ depth + (1 | id)

5

199.08

101.54

 ~ ice + (1 | id)

5

874.96

439.48

 ~ ice + (ice | id)

7

875.14

437.57

 ~ 1 + (1 | id)

5

950.24

477.12

  1. Separate models were applied to winter and summer. Model ranking is based on changes in the Akaike information criteria (ΔAIC) and likelihood ratios (ΔLR). The most parsimonious models, indicated in bold, show the AIC (not ΔAIC) and LR (not ΔLR) values. Models that did not converge are now shown.