1254394697

104

ycar using a GLM testing for effect of the period, sex and the interaction term sex*period. Then, to control for the effect of metabolically active tissues, we investigated the variation in mass-independent metabolic rates by including lean dry mass as covariate in the models.

4.4.4.2 Seasonal variation in organ mass

We used a GLM testing for the effects of period, sex and the interaction term period*sex to investigate ho w body composition varied through time. Since the mass of organs may vary with individual body size, we also included size as covariate in these analyses. We also considered the effect of storage time in the freezer. However, this effect was not significant and we did not include this variable in further models.

4.4.4.3 Relationship between metabolic performance and organ mass

Our statistical approach (see below) was limited by our sample size in the number of variables that can be entered in models. For this reason and to avoid collinearity between organ variables, we combined the mass of organs according to their functioń and seasonal variation pattems (figurę 4.1). We used Pearson correlations to test for collinearity between these organ groups and sińce all coefficients were < 0.60 (table 4.1), we included all organ groups in further analyses. Combined organs were gizzard and intestines (thereafter called “digestive organs”); heart and lungs (thereafter called “cardiopulmonary organs”); skeletal, leg, pectoralis and supracoracoideus muscles (thereafter called “muscles”); kidneys and liver (thereafter called “excretory organs'*). The brain and skin were not included in any group and were used as is.