Nest initiation and flooding in response to season and semi-lunar spring tides in a ground-nesting shorebird

Table 1 Temporal variation in nest flooding

		Response – flooded (0 = no, 1 = yes)	Gaussian			Binomial
Model	Effect type	Effect	Estimate	95% CI		Estimate	95% CI
Complex	Fixed	Intercept	0.155	−0.005	0.317	−9.3	−13.4	−5.2
		Nest latitude	−0.078	−0.099	−0.055	−2.8	−4.4	−1.2
		Spring tide cycle number	0.164	0.052	0.275	4.9	3	6.9
		Cos (Day of spring tide cycle)	−0.01	−0.038	0.017	−0.7	−1.8	0.4
		Sin (Day of spring tide cycle)	−0.03	−0.058	−0.002	−1.2	−2.2	−0.1
	Random (variance)	Year (intercept)	17%			100%
	Random (variance)	Spring tide cycle number (intercept)	33%			0%
		Female ID (intercept)	1%
		Residual	49%

The posterior estimates (medians) of the effect sizes with the 95% CIs derived from a posterior distribution of 5000 simulated values generated by the ‘sim’ function in R. Variance components were estimated by the ‘lmer’ function in R. To account for non-independence of data points, ‘Female ID’, ‘Year’ and ‘Spring tide cycle number’ (i.e. time within the laying season) were fitted as random intercepts. ‘Spring tide cycle number’ is standardized within the year so that the first spring tide cycle in the given year corresponds to the cycle when the first nest was initiated. This variable, as well as ‘Nest latitude’, were z-transformed (mean-centred and divided by standard deviation). ‘Day of spring tide cycle’ was transformed to radians (2 * number of days after the last spring tide * π/length of the given spring tide cycle [~ 14.75]) and fitted as sine and cosine of radians. Note that despite violating some model assumptions our Gaussian model fits the data better and, unlike our binomial model, also accounts for spatial auto-correlation in residuals. The binomial model lacks female identity as random intercept because models with female identity did not converge

ISSN: 1742-9994