Various approaches were applied to estimate the relationship between the phylogenetic trees of primate malaria parasites and their hosts. These analyses unanimously suggested that the evolution of the focal parasites is constrained by the phylogenetic relatedness of their hosts, and that host switches play central role in shaping the distribution of malaria species across their primate hosts. First, parasites cannot randomly infect all geographically available primates, because they appear only to be able to fit their reproduction to closely related hosts that belong to the same family (the extremely generalist P. brasilianum, and P. knowlesi that has recently established natural infection in human should be considered as exceptions). Second, event-based reconciliation of parasites to host trees suggests that host switching is the most obvious event that shapes the association between host-parasites phylogenies, but depending on the parasite phylogeny considered, the role of either co-speciation or duplication cannot be generally excluded. Moreover, the contribution of particular host-parasite associations in mediating concordance between trees is different, as some links appear to represent stronger phylogenetic structure than others. Third, the reconstruction of ancestral states of host use revealed that several switches of hosts across large phylogenetic distances might have occurred during the long evolution of primate malarias.
Patterns of co-occurrence of primate malaria parasites and hosts were comparable to those that have been observed in avian plasmodia at the family and the clade levels [9
]. Therefore, malaria parasites of vertebrates are generally conservative, and do not infect all potential hosts across wide phylogenetic distances. Only a single, highly generalist parasite, P. brasilianum
is able to infect a large number of hosts, spanning three primate families (Aotidae, Atelidae, Cebidae). The evolutionary success of this species in various hosts may result from the fact that it is able to transmit via several vector species and from the absence of other competitor parasites in South America [34
]. Moreover, it is also possible that the morphospecies P. brasilianum
consists of several genetically distinguishable lineages that specialize on different hosts.
Malaria parasites usually have a host range that is constrained by their geographic distribution. Primate hosts are not randomly distributed across the globe, as certain taxonomic groups are associated with certain localities (i.e. prosimians in Madagascar, platyrrhins in South America, and catarrhines in Africa and Asia). Hence, malaria parasites found on different continents and countries are given a choice to infect hosts that are phylogenetically related to each other. Moreover, these parasites do not randomly infect all available primate hosts, thus they seem to have limited ability to parasitize on all sympatric primate populations. This non-random selection of hosts may imply that Plasmodium species are constrained to adapt to hosts that provide similar growth environments. Given that the environment for the parasites is established by the physiological characteristics of primate hosts that are strongly determined by their phylogenetic relationships, it was predicted that the host range of parasites covers sister species that constitute the most optimal conditions for reproduction. There was a non-significant tendency for that parasites – by infecting geographically and evolutionarily related species – are faced with consistent selective pressures, as variation in host erythrocyte size tended to be smaller within the host range of parasites than across hosts of different parasites. Based on the current data of limited availability, therefore, it would be immature to exclude the scenario that each Plasmodium parasite has to cope with a more or less consistent haematological profile when adapting to a certain range of hosts. However, other factors should also be considered, as various physiological and life history characteristics can affect the host choice of malaria parasites. Accordingly, similarity in the host environment can be manifested across multiple host traits, which warrants further investigation. In any case, the range of natural hosts should reflect the outcome of a long evolutionary adaptation and thus the role of specialization that is dependent on host phylogeny.
Further analyses demonstrated that host ranges of primate malarias observed today in nature resulted from long evolutionary history of adaptation, which is shaped by host switches. Studies on various host – parasite model systems have repeatedly demonstrated that co-evolution plays a key role in shaping the resemblance of parasite phylogenetic trees to that of their hosts, while host switching remains relatively infrequent in other parasites [e.g. [61
]]. Malaria pathogens differ from ectoparasitic or bacterial parasites by being transmitted by dipteran vectors, which enhance lateral transfer between distantly related hosts and thus increase the probability of host switching [10
]. Accordingly, several host switching events can be reconstructed on the available phylogeny of primate plasmodia.
For example, there may have been an early split of cercopithecid and hominoid/platyrrhine malarias. The cercopithecid line may have successfully radiated first in Africa in ancient mandrills, and later, a more recent adaptive radiation may have occurred in Asia in ancient macaques. The presence of suitable vectors and divergent host populations are generally envisaged as the responsible factors that allowed the successful spread of parasites in this region [33
]. Moreover, even within the apparently host-conservative cercopithecid line, there are several examples of lateral transfer of malaria across hosts from different families (cercopithecid-hominoid host switch) (Figure and Additional File 3
). For example, the ancestor of the gibbon malaria, P. hylobati
derived from Plasmodium inui
-like parasite, whereas the human P. vivax
also originates from a macaque parasite. In addition, the recently gained ability of P. knowlesi
to grow in human hosts reflects the host switch of a parasite that primarily evolved to infect cercopithecid species. Furthermore, other recent nodes also involve flexible host ranges, as the ancestors of malarias infecting platyrrhine primates were probably omnipotent parasites with the capacity to grow in hominoids.
Parasites of the cercopithecid lineage demonstrate noticeable diversity in terms of life history, as their host specialization does not inevitably require the development of exclusive reproductive strategies that would restrict their potential to exploit alternative hosts. This group includes both relapsing and non-relapsing parasites, spans quartan, quotidian and tertian species and is basically represented by generalist plasmodia that can grow in up to eleven primate hosts (mean ± SE host range: 4.0 ± 0.99 species). A great diversity in fundamental reproductive strategies has thus been preserved. Such diversity can maintain the genetic variation on an evolutionary time scale, and can underlie successful host switch once conditions for transmission are fulfilled. Additionally, the associated anopheline vector populations deliver repeated chances of contacts with diverse primate fauna providing frequent opportunities for parasite transfer [34
]. Macaque malarias rely on mosquitoes from the Leucosphyrus
group that feed on various primates across broad taxonomic ranges including humans and orang-utans.
The results yield some additional details concerning the evolution of human malarias. For example, the current resolution suggests that the ascendant of P. falciparum
and Plasmodium reichenowi
may have involved species that could parasitize the ancestors of New World primates. This may imply that i) in the past, very generalist parasites may have existed that could realize host switches across large distances; or ii) that parasite lines followed the evolutionary splits of their hosts [21
]. Unfortunately, the current results cannot bring us closer to the events that occurred within
line. It thus remains plausible that the recent host ranges of these Plasmodium
species derive from a long-term co-evolution with the host species after a very early split, or from more recent switches between ape and human hosts. Moreover, the estimated ancestral states are consistent with both the anthroponosis and zoonosis theories of origin that may have been operated in South-America [22
]. It is equally likely that the ancestors of the simian parasites (P. brasilianum
and P. simium
) were human malaria species (P. malariae
and P. vivax
), or vice versa
. In order to understand the origin of New World malarias, we have to rely on the interpretations of events in association with temporal and spatial constraints [18
]. This approach suggests, at least in the case of the simium/vivax
pair that the human malaria P. vivax
originates from a Plasmodium
species infecting Asian cercopithecids that was subsequently brought to the New World and colonized simian primates giving birth to P. simium
. Besides the unresolved uncertainties, one striking pattern emerges from the ParaFit
results: the evolution of human malarias is not structured by host phylogeny, and it is likely that the history of human parasites involve host switches across large phylogenetic distances (Figure ). Consequently, the emergence of a new human pathogen cannot be predicted from the parasite phylogeny. This is likely to be explained by the preserved flexibility of malaria parasites to infect alternative hosts, and the potential to use various vector species.
Host specialization can involve some costs, as long-term commitment to a particular host may reduce the genetic variation needed for the use of alternative hosts. Therefore, extreme specialization may represent an evolutionary-dead-end, in which further evolution via host switching is limited [67
]. Accordingly, host-specific parasite lineages should be established during the early evolutionary history of the lineage, and such ancestral specialization should determine subsequent host diversification by constraining the frequency of host switches across wide phylogenetic distances [17
]. The deterministic evolution of host ranges is evident from the phylogenetic history of Plasmodium
species at a broad scale, as the specialization to the major vertebrate host groups has an ancient origin and subsequent host switching occurred rarely [11
]. However, analyses within the avian parasite clade showed that host switching occurs relatively frequently even at the subterminal nodes of the parasite phylogeny, although these switches bridge closely related hosts [10
]. On the other hand, the primate malaria system demonstrates clear host switches across large phylogenetic distances at both shallow and deep nodes of the phylogeny (Figure and Additional File 3
). This indicates that host specialization does not necessarily lead to evolutionary-dead-ends at a fine scale, and that the emergence of new diseases via lateral transfer across distantly related primate hosts cannot be ruled out.
Inferences about evolutionary events of parasites in the light of host phylogenies are sensitive to the phylogenetic information at hand. Although Bayesian approaches were used to deal with phylogenetic uncertainties, for some species no genetic data were available making it impossible to place them on the phylogenetic tree of primate malarias. Most importantly, information about the phylogenetic position of lemur (e.g. Plasmodium girardi) and hominoid (e.g. Plasmodium rodhaini, Plasmodium schwetzi and Plasmodium pitheci) parasites is lacking. Moreover, it might be that due to incomplete sampling, some lineages have not yet been discovered. By adding these missing or newly discovered parasite species to the parasite tree our picture about the details of co-speciation or host switching may change. To quantify the risk of the emergence of new malaria disease that may threaten humans or species with great concern, it would be crucial to test whether the remaining hominoid parasites are grouped together with the falciparum/reichenowi clade, or are spread throughout the phylogenetic tree. From such information, the likelihood of host switches between hosts closely related to humans could be estimated. However, from an evolutionary point of view, existing data already show strong evidence for congruence between host and parasite phylogenies and that host switching plays a key role in the evolution of malaria parasites of primates including those that infect humans.