Linear Models
Linear models were constructed to test the hypothesis that flight training (H1), dietary fat (H2), and dietary antioxidants (H3) influenced antioxidant capacity and lipid damage in the plasma at discrete timepoints (i.e., BG, PF, AF, RC), over the course of the experiment (calculated as RC-BG), during an acute flight (calculated as AF-PF), and in the liver and pectoralis at Recovery sampling. We used a global model without interaction terms that best matched this hypothesis and included possible explanatory covariates (i.e., cohort, sex, and wing chord), however, these covariates were not included in the global linear models.
To test the hypothesis that flight-training, dietary fat, and dietary antioxidants had an interactive effect on oxidative status, we compared our global models to models including a 3-way interaction between dietary fat, antioxidants, and training treatment. These models also tested the 2-way interactions between covariates. The models with the 3-way interactions were not among the best fit models, thus we report results for only the main effects.
To determine if energy expenditure during flight influenced antioxidant capacity or oxidative damage, we included energy consumed (kJ/min) during the acute flight as a fixed effect in linear models that compared both plasma oxidative status measured After-flight and the change in these measures during the acute flight (calculated as AF-PF) among diet groups across timepoints. The best fit models for AF oxidative status included energy expenditure, whereas those for change in oxidative status during the acute flight included energy expenditure x diet interaction term.