Table 2 Model selection table for the density models fitted in the second modeling step. For clarity, models containing variables with 85% confidence intervals included zero were removed30.

From: Large-scale variation in density of an aquatic ecosystem indicator species

Density

Detection

\({\boldsymbol{ {\mathcal L} }}\)

np

AIC

ΔAIC

AICω

\({\bf{A}}{\bf{I}}{{\bf{C}}}_{{\boldsymbol{\omega }}}^{{\boldsymbol{+}}}\)

D(river)

p(session + visit)

3639.93

12

7303.85

0.00

0.63

0.63

D(river)

p(session + visit + sex)

3639.62

13

7305.23

1.38

0.31

0.94

D(river:d2stem)

p(session + visit)

3641.80

13

7309.60

5.75

0.04

0.97

D(river:d2stem)

p(session + visit + sex)

3641.51

14

7311.01

7.16

0.02

0.99

D(d2urban)

p(session + visit)

3644.97

12

7313.94

10.09

0.00

1.00

D(·)

p(session + visit)

3646.70

11

7315.40

11.55

0.00

1.00

D(d2urban)

p(session + visit + sex)

3644.73

13

7315.45

11.60

0.00

1.00

D(·)

p(session + visit + sex)

3646.46

12

7316.91

13.06

0.00

1.00

  1. The model table shows the ‘Density’ and ‘Detection’ model structures, and the associated log-likelihood (\( {\mathcal L} \)), number of parameters (np), AIC values, AIC differences (ΔAIC), model specific AIC model weights (AICω), and finally the cumulative AIC model weights for each model (\({{\rm{AIC}}}_{\omega }^{+}\)). All models had the same σ and model structure (σ ~ Sex). Models are ranked by AIC, lower AIC is more supported, and ΔAIC is the difference between each model and the model with the lowest AIC value.
Back to article page