Table 2. Correlates of amphibian decline risk.
| Model | k | LL | ΔBIC | wBIC | %DE | Δ%DE |
| (a) Ecology/life-history | ||||||
| BS+RG+RG2+HB+SS+RC+RS+PC+FT | 21 | −1598.536 | 0.000 | 0.951 | 17.01 | |
| RG+RG2 | 8 | −1642.173 | 6.314 | 0.040 | 14.75 | |
| BS+RG+RG2 | 9 | −1640.552 | 9.346 | 0.009 | 14.83 | |
| BS+RG+HB+RC+RS | 14 | −1644.806 | 49.419 | <0.001 | 14.61 | |
| BS+RG+HB+RC | 12 | −1654.167 | 55.633 | <0.001 | 14.13 | |
| (b) Environmental context | ||||||
| lhb…+TM+PV+HL | 24 | −1525.251 | 0.000 | 0.934 | 20.26 | 3.25 |
| lhb…+TM+TV+PV+HL | 25 | −1525.143 | 5.954 | 0.048 | 20.26 | 3.25 |
| lhb…+TM+PM+PV+HL | 25 | −1526.291 | 8.083 | 0.016 | 20.20 | 3.19 |
| lhb…+PM+PV+HL | 24 | −1531.849 | 13.502 | 0.001 | 19.91 | 2.90 |
| lhb…+TV+PV+HL+HD | 25 | −1530.015 | 15.776 | <0.001 | 20.01 | 3.00 |
The five most parsimonious generalized linear mixed-effects models investigating (a) life history correlates of decline risk (n = 3,045) and (b) environmental context, after accounting for effects of life history correlates (top-ranked ecology/life-history model denoted as ‘lhb’–life-history base) (n = 3,121). Models include nested (hierarchical) taxonomic (Order/Family) random intercepts and geographic distance random slopes to account for spatial autocorrelation. Models were ranked according to the Bayesian Information Criterion (BIC). For ecology/life history models, the five most highly BIC-ranked models accounted for >99 % of the posterior model weight (wBIC) of the total of 40 models considered. For environmental context, model weights were more evenly distributed among the 5 most highly ranked of the 75 models considered. Terms shown are RG = range (km2), BS = body size, HB = habit, RC = reproductive cycle, RS = reproductive strategy, PC = presence/absence of parental care, SS = spawning site and FT = fertilization type, TM = mean temperature, PR = precipitation range, PM = mean precipitation, TR = temperature range, HL = % habitat lost, HD = human density (people/km2) Also shown are number of parameters (k), maximum log-likelihood (LL), difference in BIC for each model from the most parsimonious model (ΔBIC), model weight (wBIC), percent deviance explained (%DE) in the response variable (decline probability) by the model under consideration, and the difference between the %DE for the current environmental context model and the life history base (lhb) model (Δ%DE).