Skip to main content
. 2019 Oct 28;9:15391. doi: 10.1038/s41598-019-50931-5

Table 1.

Logistic regression models describing predictors of bobcat road mortality in Ohio during the years 1978–2017.

Model Variables K AICc ΔAICc AICcWt AUC
NLCD + TR + RT × DN 16 1775.59 0 0.4 0.8576
NLCD + TR + DN + RT × FR 16 1776.33 0.74 0.27 0.8603
NLCD + TR + DN 11 1777.01 1.42 0.2 0.8568
NLCD + TR + DN + RT × FR × DN 21 1777.8 2.21 0.13 0.8603
DN + TR + RT × FR 14 1795.67 20.08 0 0.8551
TR + RT × DN 13 1846.28 70.69 0 0.8462
RT × DN 12 1849.96 74.37 0 0.847
TR + DN 8 1853.47 77.88 0 0.8446
NLCD + TR + RT × FR 15 1875.3 99.71 0 0.8301
RT × FR 12 1881.08 105.49 0 0.833
TR + RT × FR 13 1882.64 107.05 0 0.8334
NLCD + TR 10 1883.45 107.86 0 0.8286
TR 7 1917.19 141.6 0 0.8145
NLCD + DN + RT × FR 10 2201.26 425.67 0 0.7538
NLCD + RT × DN 1 0 2210.68 435.09 0 0.7327
NLCD + DN 5 2233.9 458.31 0 0.7269
DN + RT × FR 8 2245.1 469.51 0 0.7456
DN + RT × DN 7 2280 504.41 0 0.6848
DN 2 2375.37 599.78 0 0.7145
NLCD + RT × FR 9 2483.38 707.79 0 0.6407
NLCD 4 2528.78 753.19 0 0.4994
Null 1 2546.96 771.37 0 0.5

Model statistics include the number of parameters per model (K), the Akaike Information Criteria score corrected for small sample size (AICc), the difference in the AICc score from the best-supported model (∆AICc), the explanatory value of each model (AICcWt), and the Area Under the Curve (AUC), denoting the predictive capability of each model. Model variables include: binned land cover types including forest, open land, and development (NLCD); road traits including the number of lanes and route type (TR); and the density of road per 1000 m buffer (DN). Interaction terms include the route type designation of township, municipal, state, US, or interstate (RT), surrounding forest cover (FR), and surrounding road density (DN). Best-supported models (those within two AICc units of the top model) are italicized.