Table 1.

Temporal trends in phenological indices (initiation date i.e., average lay date of first attempts, renesting date, average lay date, termination index, and breeding season length index), indices of predation timing (median predation date, median predation date of first attempts) and indices of predation intensity (proportion of nests predated, Mayfield estimate of predation risk), in the Rivelin Valley, Sheffield (1995–2011)

Index	Linear trend (days ± 1SE)	r2	Fdf	P-value	Linear model ΔAICcnull
Initiation date	−0.10 ± 0.28	0.01	0.141,14	0.72	+3.02
Renesting date*	−0.66 ± 0.21	0.44	10.071,13	0.007	−5.42
Average lay date	−0.12 ± 0.24	0.02	0.261,14	0.62	+2.89
Termination index*	−0.97 ± 0.15	0.77	42.501,13	<0.0001	−18.59
Breeding season length index*	−0.51 ± 0.22	0.30	5.471,13	0.04	−2.09
Predation date	0.03 ± 0.37	<0.01	0.011,14	0.93	+3.17
Predation date of 1st attempts	0.07 ± 0.46	<0.01	0.021,14	0.88	+3.16
Proportion of nests predated	−0.007 ± 0.005	0.14	2.281,14	0.15	+0.77
Mayfield estimate of predation	0.008 ± 0.005	0.15	2.481,14	0.14	+0.58

Linear and quadratic models were compared for each response variable by assessing the change in Akaike's Information Criterion corrected for small sample size (AICc) with respect to the null model, where a negative ΔAICc_null indicates evidence of a temporal trend; linear models were always more parsimonious than quadratic ones, and linear trends are therefore displayed with associated statistics.

^*denotes variables showing evidence of a temporal trend (P < 0.05 and negative ΔAICc_null).