Skip to main content

Table 5 Model selection results for the linear models of the DIN

From: Masting by beech trees predicts the risk of Lyme disease

Rank Model structure df logLik AIC ΔAIC Weight1 Weight2 r2
1 DIN ~ S + Y + B + RH1 + S:Y 11 8.1 12.3 0.0 47.0 47.0 82.5
2 DIN ~ S + Y + RLB + PRy−1 + S:Y 11 7.2 14.2 1.8 19.0 66.0 81.9
3 DIN ~ S + Y + B + SD1 + S:Y 11 6.9 14.8 2.5 14.0 80.0 81.7
4 DIN ~ S + Y + B + RH1 8 0.7 17.9 5.6 3.0 83.0 78.3
5 DIN ~ S + Y + B + S:Y 11 4.9 18.8 6.5 2.0 85.0 80.2
6 DIN ~ Y + B + SD1 8 0.2 19.0 6.7 2.0 87.0 77.9
7 DIN ~ S + Y + RLB + T1y−1 + S:Y 11 4.6 19.3 7.0 1.0 88.0 80.0
8 DIN ~ S + Y + RLB + PR + S:Y 11 4.1 20.3 8.0 1.0 89.0 79.6
9 DIN ~ S + Y + RLB + RH1y−1 + S:Y 11 4.1 20.4 8.1 1.0 90.0 79.6
10 DIN ~ S + Y + RLB + S:Y 10 2.3 20.7 8.4 1.0 91.0 78.7
11 DIN ~ S+Y + RLB + SD1 + S:Y 11 3.9 20.8 8.4 1.0 92.0 79.5
12 DIN ~ S + Y + RLB + T1y−2 11 3.8 21.0 8.7 1.0 93.0 79.4
13 DIN ~ S + Y + RLB + S:Y + S:RLB 13 7.2 21.1 8.8 1.0 94.0 81.0
14 DIN ~ S + Y + RLB + RH1 + S:Y 11 3.6 21.3 9.0 1.0 95.0 79.2
  1. Model selection results are shown for the linear models (LMs) with normal errors of the log10-transformed DIN response variable. The explanatory variables were site, year, beech masting index 2 years prior, RLB time lag, and the climate variables obtained from the weather stations and the field. The models are ranked according to their AIC. Of the 314 models in the set, only the 14 top models are shown for which the cumulative support (Weight2) is 95%. Shown for each model are the model rank (Rank), model structure (see Table 1 for the acronyms of the explanatory variables), model degrees of freedom (df), log-likelihood (logLik), AIC, difference in the AIC value from the top model (ΔAIC), model weight (Weight1), cumulative model weight (Weight2), and adjusted r-squared value (r2). The results from the full model selection are shown in Additional file 1: Section 6