Geographical variation in the 'bottom-up' control of diversity: fleas and their small mammalian hosts

Aim We searched for signs of the ‘bottom‐up’ diversity effect in the association between fleas (Siphonaptera) and their small mammalian hosts (Rodentia, Insectivora and Lagomorpha). We asked (1) whether a strong dependence of flea species richness on host species richness is characteristic for both Palaeoarctic and Nearctic realms; (2) if yes, whether the ratio of host species per flea species along the host diversity gradient is similar between the Palaeoarctic and Nearctic; and (3) whether factors other than host species richness (i.e. geographical position, climate and landscape) might better explain variation in flea species richness than host species richness. Location The study used previously published data on species richness of fleas and their small mammalian hosts from 26 Palaeoarctic and 19 Nearctic regions. Methods We regressed the number of flea species on the number of small mammal species across regions, separately for Palaeoarctic and Nearctic realms, using both non‐transformed data as well as data corrected for the confounding effects of host sampling effort and sampling area. To test whether flea species richness is determined by external factors unrelated to the host, we used stepwise multiple regressions of flea species richness against host species richness and parameters describing the geographical position, climate and relief of a region. Results When non‐transformed data were analysed, flea species richness was positively correlated with host species richness in both the Palaeoarctic and Nearctic, although the slopes of the two regressions differed significantly. After removal of the confounding effects of host sampling effort and sampling area, Palaeoarctic flea species richness remained strongly positively correlated with host species richness, whereas in the Nearctic, flea species richness appeared to be completely independent of host species richness. Results of the multiple regressions using corrected data demonstrated that in the Palaeoarctic, flea species richness was correlated with both the number of host species and the mean altitude of the region, whereas in the Nearctic, flea species richness only tended to be weakly correlated with latitude (however, this correlation turned out to be non‐significant after Bonferroni correction). Main conclusions We found evidence of bottom‐up control of flea diversity in the Palaeoarctic regions only, and not in the Nearctic. We explore several potential explanations for the different patterns observed in the two biogeographical realms, including differences in (1) levels of host specialization, (2) history of host–parasite associations and (3) landscape effects on flea diversification. We conclude that these factors combine to create different macroecological patterns in different biogeographical realms, and that diversity is not governed by the same forces everywhere.