Table 1 Generalized least square model of log-transformed distance between the used area and the preferred area during active and roosting periods.

From: Ecological Responses to Extreme Flooding Events: A Case Study with a Reintroduced Bird

Period Variable Coefficient Lower 95% CI Upper 95% CI P-value
Active Intercept 0.11 −1.26 1.47 0.87
Flood Extent 0.99 0.42 1.56 <0.001
Week −0.34 −0.67 −0.010 0.045
Week2 0.014 −0.0015 0.031 0.075
Winter −0.017 −1.88 1.85 0.95
Corr. struct: Week/Winter 0.37 0.10 0.74 NA
Roosting Intercept −0.18 −1.24 0.88 0.73
Flood Extent 0.98 0.62 0.98 <0.001
Week −0.035 −0.096 0.025 0.24
Winter 0.32 −0.62 1.27 0.49
Corr. struct: Week/Winter 0.31 −0.075 0.61 NA
Var. function: 2nd Winter 0.52 0.33 0.81 NA
  1. We defined ‘preferred’ feeding areas as those which were used by cranes in winter 2014–15, when most of the area was unflooded. Note that a p value cannot be calculated for the temporal autocorrelation structure (Corr. struct). We used leave-one-out cross validation (LOO-CV) scores to select the best model, including as predictors the weekly extent of flooding, the week in the winter, and winter. In the active period, based on model LOO-CV scores, the fixed effect was not included in our best fitting model but its coefficient is reported here for completeness. In the roosting period, the variance function allows within-group variance to differ between years. In this case the reported coefficient for the second winter represents the ratio between the standard deviation in the second winter relative to that in the first winter. Based on model LOO-CV scores week and winter were not included in the best model but coefficients are presented here for completeness. N = 44 observations.