3.2 | The body condition change pattern in winter
Based on the validation test of body condition indices, body mass (R2 = 0.5925, p-value < 0.05), fat score (R2 = 0.7206, p-value < 0.01), and the ratio of body mass/ bill + tarsus length (R2 = 0.8299, p-value < 0.001) were finally chosen as dependent variables for further analysis (detailed information in supplementary material table S1).
The regression analysis of the ordinal day effect on the body condition indices showed that both body mass (β = 0.011 ± 0.004, p < 0.01) and fat score (β = 0.012±0.004, p < 0.01) had significant positive regression on the ordinal day (Figure. 2A-2B). However, the ordinal day showed a slightly positive effect on the body mass ratio (β = 0.0002 ± 0.0001, p = 0.0897). The comparison test of body condition indices among different winter stages showed that only fat score had a significant variance among different winter stages. Fat score was typically higher in late winter as compared to pre-winter (p = 0.0028) and early winter (p = 0.0079) based on the Kruskal-Wallis test (Figure. 2C). However, the ANOVA-test of body mass and body mass ratio in any paired winter stage group showed no significant differences.