Ricklefs and Fallon  found that closely related haemosporidian parasites are conservatively distributed within host higher taxa, suggesting that codivergence or switching among closely related hosts is a common event in their evolutionary history [2, 54]. In addition, duplication (within host speciation) events were modeled by TreeFitter as being frequent among these parasites when the event cost was set low . By using the same gene (mtDNA cyt b) from previous studies, we have found that host switching is a commonly observed mechanism between haemosporidians and their non-passerine hosts, mainly close to the tips of the host phylogenetic tree (i.e., between close relatives). Codivergence is also observed in the evolutionary history of haemosporidian parasites infecting non-passerine birds, in particular at the family level (Figure 1). Duplication (within-host parasite lineage splitting) has also played some role in the diversification of non-passerine Haemosporida, which was also recognized in haemosporidian parasites infecting passerine birds under low event costs . Even when codivergence is a prominent feature of our system, strict host specificity at the species level is rare. These findings suggest a dynamic recent evolutionary history between haemosporidians and their bird hosts, which imply that host switching, in combination with geographic isolation, might be an important mechanism in the formation of new haemosporidian species (Ricklefs et al., unpublished). Cospeciation studies in fish-monogenean systems have also identified host switching as an important diversification mechanism aided by subsequent host isolation, producing a higher degree of specificity at the family level or above [55, 56]. This suggests that the evolutionary history of these parasite systems is dominated by traits (e.g., immunology of hosts) that are conserved at high taxonomic levels (i.e., family or above). Where ecological barriers do not prevent parasites from switching hosts, related parasite lineages can infect distantly related hosts, as in the case of closely related Haemoproteus parasite lineages (H. multipigmentatus, STG14, NZB9) infecting distantly related sea birds (e.g., Sula granti, Creagus furcatus) and doves (e.g. Zenaida galapagoensis), which occur on the same islands . In the absence of such opportunities, resulting from behavioral or geographic host isolation, co-speciation might become more likely [see [55, 56] for the case of fish-monogenean systems].
It is important to consider that we are leaving aside the vectors (Diptera) in this analysis of host-parasite interactions. How do haemosporidians relate coevolutionarily to dipteran vectors? It is a question for which we have almost no insight. Previous work suggests that different dipteran families are specialized in transmitting different Haemosporida genera [e.g., Ceratopogonidae transmit only Haemoproteus (Parahaemoproteus) parasites, [22, 47]. Recent work on Culicidae [57–59] and Ceratopogonidae [18, 19, 60, 61] shows that many vector species have broad feeding preferences, even across vertebrate classes. Thus, vectors may come in contact with a diverse array of Haemosporida parasites, including genera they do not normally transmit [e.g., [18, 60, 62–64]. The fact that many vector species have broad vertebrate host preferences and are susceptible to infections by different Haemosporida genera, would suggest that specificities of both vertebrate and insect host immune systems would mediate parasite “jumps” across distantly related avian hosts [see [15, 64]. Experimental studies have demonstrated that an avian parasite, Plasmodium lophurae, can adapt to and be viable in mice after just four rounds of infectious inoculations . Furthermore, erythrocytes of different mammal species have been shown to be susceptible to invasion by bird Plasmodium parasites . Hence, these studies suggest that haemosporidian parasites could potentially adapt to phylogenetically distant hosts.
Parasites of the sub-genus Haemoproteus, which are normally transmitted by louse flies (Diptera: Hippoboscidae), might be less likely to switch between unrelated hosts because normally hippoboscid flies are host specific and do not fly long distances . However, recent work in the Galapagos Islands has shown that lineages belonging to the same morphospecies (H. multipigmentatus) infect endemic doves [24, 28] and sea birds inhabiting the same islands [frigate birds, gulls, brown boobies, , indicating host sharing by the flies even though different-sized hippoboscid species are associated with these different bird species. Furthermore, hippoboscid flies (Olfersia spinifera and O. aenescens) have higher rates of gene flow in comparison the their seabird hosts, which might explain the lack of genetic structure of the Haemoproteus iwa parasite across the Galápagos Islands . Finally, the evolutionary history of the Hippoboscidae includes at least two host switches from mammals to birds , suggesting that parasites either hitchhiking with louse flies [e.g. phoretic mites,  or developing within them (e.g. Haemoproteus) can “jump” across large host phylogenetic distances. Hence, under the right ecological conditions (e.g. simple systems like island faunas, different bat species sharing roosts , similar parental care behaviour across different cichlid species ) parasites, in particular haemosporidians transmitted by hippoboscid flies, are capable of switching across distantly related hosts (see Figure 1).
Our vertebrate host sample includes species that belong to the same genera (e.g., Columbina spp., Zenaida spp.) within the same host sub-family (e.g., Columbidae: Columbinae). The distribution of parasites among host species reported here supports a degree of host specificity below the family level for parasite lineages (e.g., lineages of H. multipigmentatus, H. iwa, H. (Parahaemoproteus) syrnii, see Figure 1), which is validated by the cospeciation events and the global cospeciation signal detected in our system. However, that parasite haplotypes of H. (Haemoproteus) infect many species and genera within Columbiformes , sea birds of the same genus e.g., Fregata, , and host species across orders (e.g., H. (Parahaemoproteus) FALC11, Figure 1), lends support to a diversification history dominated by host switching. Our sample consists mostly of a small number of non-passerine species of the sub-family Columbinae, the genera Fregata, Creagus, Todus, Falco, Bubo, Strix, Anas, Spheniscus, Picoides, and Alcedo along with their described haemosporidian parasites, which contrasts with analyses of widely sampled passerine hosts and their parasites [see [2, 9]. Thus, although our results support previous findings, a broader non-passerine taxonomic sampling with broader geographical distribution will be required to produce a more comprehensive understanding of haemosporidian diversification.
Previous work considered that all parasites infecting Columbiformes belonged to the H. (Haemoproteus) subgenus [24, 28]. However, in this study we identified a parasite lineage (CHI_30PA) from a Zenaida macroura dove that falls within the strongly supported H. (Parahaemoproteus) subgenus, which is in agreement with other recent examples of columbiform parasites (H. sacharovi and H. turtur) that belong to the subgenus Parahaemoproteus and likely are transmitted by Culicoides vectors .