Résumé
We studied the co-occurrence of flea species in infracommunities of 16 rodents from four regions (South Africa, Tanzania, central Europe and western Siberia) using null models, and predicted that flea co-occurrences will be expressed more strongly in male than in female hosts. We examined patterns of co-occurrence (measured as the C score) in infracommunities of fleas that are parasitic on male and female hosts by comparing co-occurrence frequencies with those expected by chance. When a significant degree of nonrandomness in flea co-occurrences was detected, it indicated aggregative infracommunity structure. In Tanzanian rodents, no significant flea co-occurrences were detected in either male or female hosts. In a South African rodent, significant flea co-occurrences were not detected in males, but were found in females in some localities. In Palaearctic rodents, significant nonrandomness was detected either equally for males and females or more frequently in males than in females. Meta-analyses demonstrated that the frequency of the detection of nonrandomness in flea co-occurrences was significantly higher in male than in female hosts. The values of the standardized effect size (SES) for the C score differed significantly among host species, but not between host genders. When the Palaearctic hosts were analyzed separately, the effects of both host gender and species appeared to be significant, with the SES values for the C score in males being smaller than those in females. The strength of the gender difference in the manifestation of flea community structure increased with increasing gender difference in flea species richness, and with decreasing gender difference in flea prevalence for the Palaearctic hosts. We conclude that male hosts are the main drivers of flea infracommunity structure. However, the manifestation of gender bias in flea community structure varies among host species, and is likely determined by the pattern of species-specific spatial behavior.