Skip to main content
. 2016 Mar 22;6:23546. doi: 10.1038/srep23546

Table 1. Effects of experimental manipulations on beak colouration, plasma carotenoids, body mass and blood redox state.

Response variables Predictors Parameter estimates Model statistics Standardised effect size
Mean effect SE t P b* CI 2.5% CI 97.5%
Beak red chroma
 initial 0.17 0.10 1.64 0.11 0.13 −0.03 0.28
 ROS −0.26 0.05 −4.81 <0.001 −0.73 −1.04 −0.43
 CAR 0.50 0.05 9.38 <0.001 1.44 1.13 1.75
 ROS × CAR 0.07 0.11 0.61 0.54 0.19 −0.43 0.80
Total plasma carotenoids
 initial 0.08 0.02 4.35 <0.001 0.31 0.17 0.45
 ROS −0.71 0.24 −2.99 0.005 −0.42 −0.70 −0.14
 CAR 2.64 0.24 11.08 <0.001 1.56 1.28 1.84
 ROS × CAR 0.29 0.48 0.62 0.54 0.17 −0.39 0.74
ZE/tHODE ratio
 initial −0.36 0.25 −1.44 0.16 −0.06 −0.15 0.02
 ROS 0.85 0.04 20.10 <0.001 1.55 1.39 1.71
 CAR −0.57 0.04 −14.88 <0.001 −1.04 −1.18 −0.90
 ROS × CAR −0.42 0.09 −4.86 <0.001 −0.77 −1.08 −0.45
OXY
 initial 0.67 0.13 5.19 <0.001 0.60 0.37 0.83
 ROS 20.56 7.29 2.82 0.007 0.64 0.19 1.11
 CAR −1.15 7.05 −0.16 0.87 −0.04 −0.48 0.41
 ROS × CAR 4.13 14.17 0.29 0.77 0.13 −0.77 1.03
8-isoprostane in RBC
 initial 0.41 0.09 4.48 <0.001 0.53 0.29 0.76
 ROS 1.97 1.04 1.89 0.07 0.43 −0.03 0.90
 CAR 1.56 1.06 1.47 0.15 0.34 −0.13 0.81
 ROS × CAR −1.35 2.12 −0.64 0.53 −0.30 −1.24 0.64
Body mass
 initial 1.01 0.11 9.62 <0.001 0.83 0.66 1.00
 ROS −0.02 0.31 −0.06 0.95 −0.01 −0.33 0.31
 CAR −0.26 0.33 −0.80 0.43 −0.14 −0.48 0.21
 ROS × CAR −1.44 0.62 −0.62 0.02 −0.75 −1.40 −0.11

Estimates represent coefficients from linear models with oxidative challenge (ROS) and carotenoid intake (CAR) included as factors and pre-treatment (initial) values as covariates. Low and high factor levels were coded 0 and 1, respectively and centred in order to enable the main effects to be properly interpreted without the need to remove the interaction terms from the models. Proportional variables (i.e. beak red chroma and ZE/tHODE ratio) and total plasma carotenoids were normalised using logit and Box-Cox (λ = 0.335) transformation, respectively. Standardised effect sizes are reported as standardised partial regression coefficients (b*) from the same models with continuous variables z-standardised.