A total of 95 females (47 medicated, 48 controls) and 92 males (47 medicated, 45 controls) were included in the experiment. At first capture, 85 per cent of females and 88 per cent of males were infected with Haemoproteus spp. (see also table S1 in the electronic supplementary material), whereas 92 per cent of females and 87 per cent of males were infected by Leucocytozoon spp. All except one male and one female were infected by at least one parasite.
Repeated-measures ANOVA revealed that medicated females experienced a significantly greater reduction in intensity of infection by Haemoproteus parasites from initial to final samples than control females (a; F1,77 = 4.25, p = 0.04). In males, change in intensity was not significantly affected by treatment (b; F1,73 = 0.34, p = 0.56). There were no significant treatment effects on change in intensity of infection by Leucocytozoon parasites in either sex (all p > 0.35).
Figure 1. Change in the intensity of infection by Haemoproteus parasites (log-transformed) in (a) female and (b) male blue tits with respect to the medication treatment (females: F1,77 = 4.25, p = 0.04; males: F1,73 = 0.34, p = 0.56). Bars denote 95% confidence (more ...)
Of all birds included in the study in 2004, 28 females (19 medicated and nine controls) and 26 males (11 medicated and 15 controls) were recaptured at least once between 2005 and 2007. Sex and treatment were not significantly associated with local survival, but there was a significant interaction effect on survival (GLZ: treatment: Wald = 0.77, p = 0.38; sex: Wald = 0.001, p = 0.97; treatment × sex interaction: Wald = 5.56, p = 0.02), supporting an effect of medication on survival in females only. Accordingly, the best model for survival included an interaction between sex and treatment in the first year post-treatment (model 1, table S2 in the electronic supplementary material). The other three competing models included within two Akaike Information Criterion with second order correction units of the best model (table S2 in the electronic supplementary material) support the interaction between sex and treatment on survival between 2004 and 2005, although differing in effect of treatment on survival to subsequent years. Furthermore, the only partial coefficient of the log-linear function for survival for the best-fit model with a 95% confidence interval that did not overlap with zero included the interaction between sex and treatment (Beta (s.e.) = 1.71 (0.60), 95% confidence interval = 0.53–2.89). This indicates that survival probability of treated females was higher than for controls, and there was no significant effect of primaquine treatment on male survival. We estimated survival probabilities for males and females assigned to different treatments by model averaging, and the higher survival probability was obtained for medicated females (). For additional information, see the electronic supplementary material.
Survival probabilities ± standard error (s.e.) and 95% confidence interval for control and medicated blue tits. (Estimation was done using model averaging.)