*Between 2013 and 2015, the leukocyte concentration was determined for 79 samples from 34 individuals sampled between two and three times. 23 individuals had two samples in different years and 11 individuals had three.

**Between 2011 and 2015, the leukocyte counts were determined for 169 blood smears from 52 individuals sampled between two and five times. 18 individuals had two samples in different years, 12 had three, 13 had four, and 9 individuals had five. 33 individuals were both measured for leukocyte concentration and leukocyte counts (which corresponds to 75 measurements).

Table 2: Parameter estimates and credible intervals at 95% (CI) for the selected models testing whether variation in leukocyte concentration or relative number of each type of leukocytes were explained by age, age at access to dominance (AAD, i.e. selective appearance) and age at last observation (ALO, i.e. selective disappearance). Parameters were obtained from the minimal adequate models. Significant effects (CI which do not overlap zero) are in bold. “-“ means a parameter not retained in the model.