Factors influencing nest-site selection
A total of 33 nests (all located in wetlands varying levels of standing
water) and an equal number of randomly selected nest-sites were accessed
and measured. Correlation between predictor variables (water depth,
vegetation height, distance to the water edge and grazing intensity)
were first considered for analysis to test for multicollinearity
(prediction with a substantial degree of accuracy). The VIF values were
low (<1.3), and therefore all predictor variables were
included in the model without correcting for multicollinearity. The
model selection algorithm yielded a subset of 8 best models, where the
most important predictor variables (i.e. those with ΔAICc <2)
were determined as i) water depth, ii) vegetation height, and iii)
distance of nest location from the water edge (Figure 2). Water depth
and vegetation height had positive effects on nesting probability while
the distance to water edge exhibited a quadratic tendency (Figure 3).