Skip to main content
. 2014 Apr 16;9(4):e95126. doi: 10.1371/journal.pone.0095126

Table 1. Summary of stepwise selection procedure for presence-absence model of right whales in the southeastern US.

Model % Deviance GCV mean ASPE
null 0 0.3417 0.0392
s(SemiMonth:Northing) 10.3 0.3065 0.0377
s(SemiMonth:Northing)+s(DistToShore) 15.7 0.2882 0.0364
s(SemiMonth:Northing)+s(DistToShore)+Year 18.8 0.2777 0.0352
s(SemiMonth:Northing)+s(DistToShore)+Year+s(Effort) 20.9 0.2706 0.0345
s(SemiMonth:Northing)+s(DistToShore)+Year+s(Effort)+s(SST) 22.3 0.2658 0.0342
s(SemiMonth:Northing)+s(DistToShore)+Year+s(Effort)+s(SST)+s(Depth) 22.6 0.2647 0.0340
s(SemiMonth:Northing)+s(DistToShore)+Year+s(Effort)+s(SST)+s(Depth)+s(DistTo22Iso) 22.8 0.2642 0.0340

Predictor variables include interaction between semimonthly period and UTM northing, distance to the shoreline (DistToShore), survey year, survey effort, sea surface temperature (SST), bottom depth, and distance to the 22°C SST isotherm (DistTo22Iso). Smoothed covariates indentified by “s()”. Evaluation criteria include the proportion of deviance explained, generalized cross validation score (GCV), and mean average squared prediction error (ASPE) from a five-fold cross-validation.