RESULTS
The beta diversity and normalized Sackin index of parasites varied according to the host-switching intensity (i.e. the expected percentage of parasite individuals that switch hosts during the entire history of the host community). To illustrate the prevalent beta diversity and normalized Sackin index according to the host-switching intensity, we presented the case of feather mites associated with birds (ID. 1 - Fig. 1). As expected, beta diversity decreases as host-switching intensity increases regardless of the studied case (Fig. 1a and Fig. S14). It occurs because the increase of host-switching promotes the interaction of different host species with the same parasite species. Additionally, for each value of host-switching intensity, there is a small variation in the beta diversity (Fig. 1a and Fig. S14). The only exception was ID. 4, which resulted in a wide variation in beta diversity under high host-switching intensity (Fig. S14).
When host-switching intensity is low (below 5%), the normalized Sackin index (In) for the simulated parasite phylogenies assumes exactly the same value of the one obtained from the empirical phylogeny of the host (note the dashed line in Fig. 1b and also Fig. S15). This low host-switching intensity does not allow the establishment of a population in a new host. As a consequence, the simulated parasite phylogenies have the same normalized Sackin index of the empirical host phylogeny. For higher host-switching intensity, colonization followed by speciation is more likely to occur and the normalized Sackin index varies over simulations even when they are under the same host-switching intensity (Fig. 1b and Fig. S15). The wide variation in the normalized Sackin index for a given host-switching intensity reveals that stochastic host-switching events, even if more likely occurring between closely related species, can change the structure of the resulting phylogenetic tree. The normalized Sackin index tends towards zero (balanced tree) as host-switching intensity goes to one, regardless of the community (Fig. S15), resembling a neutral speciation scenario Yule model (Yule 1924; Aldous 2001).
For all cases, there is a host-switching intensity that simultaneously reproduces the empirical beta diversity and the parasite normalized Sackin index (Fig. 2). As mentioned before, both metrics are sensitive to host-switching intensity, but still restricted to some combinations of beta diversity and Sackin index (see in Fig. 2, the graphs showing the beta diversity vs. normalized Sackin index are not fully filled). Generally, the beta diversity and the parasite normalized Sackin index obtained under high host-switching intensity (greater than 50%) are far from the empirical pattern (see Fig. 2, the yellow dots rarely approach the intersection of the solid lines). Then, host phylogeny is an important proxy for host-switching events.
The simulated host-switching intensity that simultaneously fit beta diversity and parasite normalized Sackin index of each empirical case (\(\pm\)5% confidence interval) recovered a range of 0.06% to 22.07% of host-switching intensity. Within this range, the associations between mammals and lice presented the lowest host-switching intensity (case ID. 2 with 0.07% - 1.13%), followed by that involving birds and feather mites (case ID. 1 with 0.06% - 8. 17%), wildlife and arthropod parasites (case ID. 3 with 0.82% - 2.69%), frogs and Polystomatidae (case ID. 7 with 0.22% - 3.71%), frogs and Rhabdias spp. (case ID. 8 with 1.99% - 4.94%), frogs and Oswaldocruzia spp. (case ID. 9 with 5.29% - 5.91%), fish and Gyrodactylidae (case ID. 6 with 9.35%), - the highest intensities of host-switching were observed between rodents and fleas (case ID. 5 with 13.47% - 16.87% and case ID. 4 with 0.43% - 22.07%). For ectoparasites, the switching intensity of hosts observed in the simulations was 0.06% to 22.07% and, for endoparasites, it was in the range of 0.22% to 5.91% but no significant correlation was observed between host-switching intensity and parasitism type (LMM: beta= -0.007, SE= 0.018, df = 5.98, t= -0.6, p = 0.70, ANOVA: F=0.16, p = 0.69, Fig. 3). The host-switching events are more frequent in studies conducted in a local scale (blue color in Fig. 3) than in regional scales (salmon color in Fig. 3) (LMM: effect of host-switching intensity on spatial scale: beta= 0.08, SE= 0.18, df = 6.12, t= 4.66, p = 0.003, ANOVA: F = 21.75, p = 0.003, Fig. 3).