Skip to main content
. 2018 Jun 5;13(6):e0197767. doi: 10.1371/journal.pone.0197767

Table 3. Generalized Additive Mixed effects models with the probability of an individual seal being hauled out in each hour of a trip modelled as a binomial response.

Individual seal was a modelled as random effect with trip nested within it. AIC values are reported for the best model fit and the effect of adding (+) or removing (-) explanatory variable including different ways of grouping smoothers. Correlation was modelled by including terms of whether the seal was hauled out in the preceding hours (lag1-5). Time of year was compared using Julian day, week of year, month.

Model terms AIC BIC deviance df.resid
Best model
Hour (by month) + Latitude (by week) + Distance to land/ice (by tag) + week (by tag) + lag1 + lag2 + lag3
10525.1 10845.9 10449.1 34222
Comparison with alternative models
+ lag4 10523.1 10853.1 10445.9 34221
-lag3 10533.7 10846.1 10459.7 34223
-distance to land/ice(by tag) 10557.8 10817.7 10497.8 34230
-'(by tag)' from 'distance to land/ice' 10546.2 10821.3 10482.2 34228
-'(by tag)' from 'week' 10539.8 10837.7 10469.8 34225
'month(by tag)' replacing 'week(by tag)' 10610.0 10930.8 10534.0 34222
'Julian day(by tag)' replacing 'week(by tag)' 10553.7 10874.5 10477.7 34222
-'(by tag)' from 'Latitude' 10550.6 10820.8 10486.6 34228
Latitude (by tag) 10873.3 11118.0 10817.3 34232
Latitude(by month) 10486.6 10929.2 10378.6 34206
hour(by Tag) 10794.6 11084.9 10726.6 34226