The phylogeography of trypanosomes from South American alligatorids and African crocodilids is consistent with the geological history of South American river basins and the transoceanic dispersal of Crocodylus at the Miocene

Background Little is known about the diversity, phylogenetic relationships, and biogeography of trypanosomes infecting non-mammalian hosts. In this study, we investigated the influence of host species and biogeography on shaping the genetic diversity, phylogenetic relationship, and distribution of trypanosomes from South American alligatorids and African crocodilids. Methods Small Subunit rRNA (SSU rRNA) and glycosomal Glyceraldehyde Phosphate Dehydrogenase (gGAPDH) genes were employed for phylogenetic inferences. Trypanosomes from crocodilians were obtained by haemoculturing. Growth behaviour, morphology, and ultrastructural features complement the molecular description of two new species strongly supported by phylogenetic analyses. Results The inferred phylogenies disclosed a strongly supported crocodilian-restricted clade comprising three subclades. The subclade T. grayi comprised the African Trypanosoma grayi from Crocodylus niloticus and tsetse flies. The subclade T. ralphi comprised alligatorid trypanosomes represented by Trypanosoma ralphi n. sp. from Melanosuchus niger, Caiman crocodilus and Caiman yacare from Brazilian river basins. T. grayi and T. ralphi were sister subclades. The basal subclade T. terena comprised alligatorid trypanosomes represented by Trypanosoma terena n. sp. from Ca. yacare sharing hosts and basins with the distantly genetic related T. ralphi. This subclade also included the trypanosome from Ca. crocodilus from the Orinoco basin in Venezuela and, unexpectedly, a trypanosome from the African crocodilian Osteolaemus tetraspis. Conclusion The close relationship between South American and African trypanosomes is consistent with paleontological evidence of recent transoceanic dispersal of Crocodylus at the Miocene/Pliocene boundaries (4–5 mya), and host-switching of trypanosomes throughout the geological configuration of South American hydrographical basins shaping the evolutionary histories of the crocodilians and their trypanosomes.


Background
There is increasing evidence that the evolutionary histories of hosts and parasites are associated, and consequently, that parasites can serve as proxies to understand host evolutionary history and vice versa. Comparative phylogeography and biogeography of hosts and obligate parasites can reveal the impressions left on contemporary species by evolutionary processes such as co-evolution, dispersion and colonisation [1][2][3][4][5][6][7].
The genus Trypanosoma (Kinetoplastea, Trypanosomatidae) comprises worldwide parasites of all vertebrate classes transmitted by hematophagous vectors, such as insects, ticks and leeches [8,9]. Trypanosomes have been described in many lizards, snakes and crocodilians [10]. The evolutionary histories of trypanosomes from mammals of distinct orders and from other vertebrate classes only recently began to be addressed [11][12][13][14][15][16][17][18][19]. Trypanosomes parasitizing alligatorids have been reported in Ca. crocodilus and Ca. yacare from Brazilian Amazonia and Pantanal, respectively [20,21]. In Africa, trypanosomes of Crocodylus niloticus, Mecistops cataphractus and Osteolaemus tetraspis were all named T. grayi [22,23], a species that has been commonly described in its vector tsetse flies [24][25][26][27]. The transmission of T. grayi to crocodiles occurs by oral contamination with tsetse faeces or the ingestion of crushed flies [23]. The South American vectors of crocodilian trypanosomes are unknown. The first trypanosomes from alligatorids established as continuous cultures were from the Brazilian Ca. yacare. Nine isolates were molecularly characterised and assigned to two genotypes, Cay01 and Cay02, distinct from T. grayi from the African C. niloticus. However, previous phylogenetic analysis restricted to a few crocodilian trypanosomes hampered any attempt to evaluate species richness, and to infer biogeographical scenarios underlining the evolutionary history of crocodilian trypanosomes [8,9,12,21].
In an evolutionary story that spans more than 200 million years, from the late Triassic to present, a very large diversity of crocodilians occupied worldwide habitats, but only a limited number of species survived the long and intricate Crocodylia evolutionary history [28][29][30][31][32][33]. The Crocodylia comprises species of Crocodylidae, inhabiting all continents, and of Alligatoridae, which are exclusive to the Western hemisphere (except for the Chinese alligator). The Alligatoridae and Crocodylidae diverged in the late Cretaceous (~90 mya) [28,32]. M. niger occurs in the Amazon basin; Ca. crocodilus is the most widespread species in the Latin America; and Ca. yacare occurs mainly in the Paraguay/Paraná basin [34]. The current limited distribution and small number of Crocodylus species in South America (C. intermedius in the Orinoco basin and C. acutus from northwest Colombia/Venezuela) contrast with great diversity and wide distribution of fossils at the Miocene (9-5 mya) [35,36]. C. niloticus is the most widespread crocodilian species in Africa, and more closely related to South American than to Asian species of Crocodylus, whereas the African Osteolaemus tetraspis together with Mecistops cataphractus constitute the sister clade of Crocodylus [29][30][31][32].
In this study, we characterized trypanosomes from the South American Ca. crocodilus, M. niger and Ca. yacare and the African C. niloticus and O. tetraspis aiming: a) to evaluate the relative forces of host species and geography in shaping the genetic diversity, emergence of new species, phylogenetic relationships and distribution of crocodilian trypanosomes; b) to hypothesise paleontological and biogeographical scenarios consistent with both trypanosome and crocodilian evolutionary histories.

Methods
Collection sites, handling of the crocodilians, and culture of trypanosomes After immobilisation, the animals were bled by tail puncture using sodium citrate as an anticoagulant. The distribution of the crocodilian species and locals within each South American hydrographic basin where the animals were captured are shown in the Figure 1a, Table 1. We examined crocodilian blood for trypanosomes under light microscopy in Giemsa-stained smears on glass slides. Blood samples (~1.0 ml) were immediately inoculated into tubes containing BAB-LIT medium, and the positive haemocultures were transferred to monolayers of Hi-5 feeder cells cultivated in TC100 medium as previously described [14,19,21]. All cultures were deposited in the Trypanosomatid Culture Collection of the University of São Paulo (TCC-USP) ( Table 1).

DNA amplification, sequencing and data analyses
DNA preparations from cultured trypanosomes were obtained using the classical phenol-chloroform method. The DNA preparations from blood samples were obtained from samples preserved in ethanol (v/v) stored in our Blood Sample Collections (BSC). DNA of T. grayi (CroCamp1) of C. niloticus from Cameroon, included in a previous study [12], was kindly provided by S. Helder, University of Yaoundé, Cameroon. PCR-amplification and sequencing of the V7V8 region of SSU rRNA (small subunit of rRNA) and glycosomal glyceraldehyde phosphate dehydrogenase (gGAPDH) genes were performed as described previously [37,38]. The sequences obtained were employed for the following phylogenetic analyses: a) V7V8 SSU rRNA sequences (880~bp) from the crocodilian trypanosomes; b) gGAPDH sequences (864 bp) from the crocodilian trypanosomes and species representing all major clades within Trypanosoma using non-trypanosome trypanosomatids as an outgroup; and c) concatenated V7V8 SSU rRNA and gGAPDH sequences (~3.3 kb) from the crocodilian trypanosomes and their closest related trypanosomes. The species included in the phylogenetic trees and their respective GenBank accession numbers are shown in in the phylogenetic trees and Table 1. The alignments were employed for Parsimony (V7V8 SSU rRNA), maximum likelihood (ML) and Bayesian inference (BI) analyses as previously described [15,19,21,37,38].

Light, Scanning (SEM) and Transmission (TEM) Electron Microscopy
For light microscopy, cultured trypanosomes were fixed with methanol and stained with Giemsa. For Scanning (SEM) and Transmission (TEM) Electron Microscopy, trypanosomes from Hi-5 cultures were prepared and observed as previously described [14,19,37,38].

Isolation in culture and growth behaviour
We collected blood from crocodilians captured in four South American river basins: PP, AM, AT and OR ( Figure 1; Table 1) and characterized 5 new cultures: two from Ca. yacare (TCC1611, 1974), two from Ca. Crocodilus (TCC 1829, 2218) and one from M. niger (TCC 1838). We observed accentuated growth differences by culture comparison. The absence of Hi-5 cells markedly reduced the multiplication of the TCC 610 and 1611 isolates, whereas the remaining isolates easily adapted to grow on TC100 or LIT media without the feeder layer cells.
One primary culture (BSC28) was obtained from Ca. crocodilus from Venezuela (Table 1). This isolate and other Brazilian isolates failed to propagate beyond the first passage in culture. Some blood samples that tested positive for trypanosomes by microscopic investigation of fresh blood could not be cultured at all. To overcome this limiting factor, DNA isolated directly from crocodilian blood samples was used for PCR amplification and sequencing of the V7V8 SSU rRNA and/or gGAPDH sequences.

Barcoding, species diversity and the phylogenetic relationships of crocodilian trypanosomes
In previous studies, we demonstrated that barcoding using V7V8 SSU rRNA sequences is valuable for rapid species identification, and for revealing intra-specific diversity of trypanosomes in mammals [16][17][18][19] anurans [15], and snakes [13,14]. We previously employed this marker to characterise the genotypes Cay01 and Cay02 in Ca. yacare [21]. Here, we barcoded two new cultures from Ca. yacare, two from Ca. crocodilus, and one from M. niger. Barcoding the trypanosomes directly from crocodilian blood samples disclosed sequences that were identical to Cay01 or Cay02 genotypes (Table 1; Figure 2).
The SSU rRNA sequences obtained from the 5 new established cultures, the primary culture derived from the Venezuelan Ca. crocodilus, and 8 sequences obtained directly from the DNA from blood samples of M. niger, Ca. yacare and Ca. crocodilus and O. tetraspis were aligned with those previously determined [8,9,12,21] for 9 isolates from Ca. yacare and three isolates of T. grayi, including the trypanosome from C. niloticus from Cameroon barcoded in this study ( Table 1). The inferred dendrogram corroborated the previous separation [21] of the alligatorid trypanosomes into the main genotypes Cay01 and Cay02, both distinct from T. grayi ( Figure 2). The barcode analysis showed that Cay01 and Cay02 diverged~2.5%, Cay02 and T. grayi also diverged~2.5%, whereas a smaller divergence (1.3%) separated Cay01 from T. grayi. The gGAPDH gene (single copy gene) sequence can be more valuable than SSU rRNA genes (multiple copies) for the phylogenetic positioning of trypanosomatids [9,14,15,17,18,37,38]. Here, we showed that among the crocodilian trypanosomes, gGAPDH diverged more than SSU rRNA. The gGAPDH+V7V8 SSU rRNA ( Figure 3) or gGAPDH ( Figure 4) derived phylogenetic trees supported the monophyly of crocodilian trypanosomes resulting in a well-supported assemblage named the Crocodilian clade. This clade comprises three well-supported subclades and confirms the branching pattern inferred by the V7V8 SSU rRNA (Figure 1). The Crocodilian clade was positioned closer to the major assemblage of trypanosomes from mammals and from Archosaurian hosts, such as birds, lizards and snakes, than to the trypanosomes from fishes and anurans nested into the Aquatic clade (Figures 3, 4).
The results strongly support a crocodilian clade that consists exclusively of and comprises all trypanosomes from crocodilians, including isolates obtained from tsetse flies (Figures 3, 4). The positioning in the phylogenetic trees and gGAPDH sequence divergences, permitted classification of the Brazilian isolates into two new species, Trypanosoma terena n. sp. and Trypanosoma ralphi n. sp. The gGAPDH divergence between T. ralphi (the isolate TCC1838 from M. niger captured in the AM basin represents the type species) and T. terena (the isolate TCC610 from C. yacare captured in PP basin represents the type species) was 7.6%. The divergence separating T. ralphi and T. grayi was 7.0%, whereas the largest distances (9.5%) separated T. terena from T. grayi.
The phylogenetic analyses supported three subclades within the clade Crocodilian. The basal subclade comprised T. terena and four isolates from South America, one from Ca. yacare, one from Ca. crocodilus and two from M. niger, and also included the trypanosome from the African O. tetrapsis. The second subclade clustered together with T. ralphi from 16 Brazilian isolates, 10 from Ca. yacare, three from Ca. crocodilus, and three from M. niger. The third subclade was formed exclusively by African isolates, which grouped T. grayi of C. niloticus from Cameroon, and two tsetse isolates from The Gambia. T. ralphi and T. grayi are sister clades (Figures 3, 4). The most homogeneous subclade (< 0.5% gGAPDH divergences) was represented by T. ralphi, and South American isolates. The only exception within this clade was the TCC 1102 isolate, which diverged by~1.0% from all isolates. Within the subclade T. terena, the isolate TCC 610 diverged by higher distances: 1.8% for the isolate from O. tetrapsis, 1.3% for the Ca. crocodilus isolate from OR, and 1.4% for the C. yacare isolate TCC1611 from PP. Considering that the divergences within the subclades are too small to justify the creation of new species, the isolates were designed as genotypes of the leading species of their respective subclades.
The African isolates from C. niloticus and O. tetrapsis were positioned into the two distant subclades T. grayi and T. terena (Figures 3, 4), and were separated by the largest gGAPDH divergences (9.6%). The divergences separating the African trypanosomes were larger than those separating T. grayi from the South American trypanosomes nested into the subclade T. ralphi (6.3%). Divergences separating the isolates of T. grayi from C. niloticus (Cameroon) and tsetse flies (The Gambia) were 1.2% (isolate BAN1) and 1.7% (isolate ANR4), whereas the two tsetse isolates diverged by 1.9%. These findings underline the heterogeneity of the African crocodilian trypanosomes, all assigned to T. grayi, which probably comprise more than one species.

Light microscopy and the development and differentiation of the new species of alligatorid trypanosomes
An examination of the Giemsa-stained blood smears revealed a small number of C-shaped and roll-shaped (rounded forms with overlapping extremities) trypomastigotes in Ca. yacare blood samples as previously observed [21]. Blood flagellates were not observed in other crocodilian species. Hemocultures in TC100 using Hi-5 as feeder cells exhibited flagellates resembling blood trypomastigotes (Figure 5a,e). After~7 days, the flagellates gradually differentiated to epimastigotes, and a few small trypomastigotes appeared in T. terena cultures (Figure 5b,f ).
The flagellates from log-cultures ranged from small rounded forms to large epimastigotes. In T. ralphi (TCC 1838), epimastigotes varied in length and were generally wide forms with bodies measuring 17.5 to 32.  (Figure 5h,i) were common in the stationary cultures. Consistently, the epimastigotes of both species exhibited segments of the flagellar membrane that were detached from the cell body membrane before the emergence of free flagellum (Figure 5c,d,i). T. ralphi exhibited a thin kinetoplast adjacent to the central nucleus (Figure 5c,d), whereas the kinetoplasts in T. terena were rounded, compacted, and positioned either close or far from the nucleus (Figure 5g,h).  Table 1), and trypanosomes from other hosts (GenBank accession numbers are within parenthesis on the tree). Numbers at nodes are bootstrap support >50% (P/ML) or Bayesian posterior probability > 0.25, derived from 500 replicates.

Scanning (SEM) and transmission electron microscopy (TEM) of cultured trypanosomes
SEM microphotographs of T. ralphi (Figure 6a,c,e,f,h,i) and T. terena (Figure 6d,g,j,k) show flagellates varying from small and rounded forms to long epimastigotes. The epimastigotes typically divide by binary fission (Figure 6j) while rounded flagellates divided by multiple fission as squash-like forms (Figure 6c,g). Epimastigotes of T. terena displayed a surface protuberance correspondent to the subjacent kinetoplast (Figure 6k). Both trypanosome species exhibited a cytostome opening, represented by a small orifice in the cell membrane close to the flagellar pocket (Figure 6f ). In addition to areas of detachment of the flagellum and the cell body membranes (Figure 6d,h,j,k), the two species exhibited highly developed flagellar lamellae (Figure 6i,j).
However, some features shared by both species were noticeable: a) large amount of organelles resembling reservosomes [19,39], which are compartments that accumulate endocytotic macromolecules ( Figure 7e); b) an apparent short invagination of the cytostome inside the cytoplasm (Figure 7g,h); c) very well developed spongiomes [37], which are a network of tubules converging to the contractile vacuole attached to the membrane of the flagellar pocket constituting the osmoregulatory apparatus (Figure 7c,d), and d) a conspicuous paraflagelar rod (PR), a network of filaments that runs alongside the flagellar microtubules (Figure 7l). The thickness and the arrangement of the kDNA fibrils differed between the two species. In T. ralphi (Figure 7i), the kinetoplast exhibits a diskshaped structure with two bands of fibrils and thickness varying from 170 to 190 nm (176.7± 6.5 nm). In T. terena (Figure 7j), the kinetoplast is highly compacted with no visible particular organization, and the thickness varied from 162 to 241 nm (198 ± 22 nm).   Figure 1 and

Taxonomical comments
Lainson [20] described Trypanosoma cecili based on flagellates observed in tissue imprints of Ca. crocodilus captured at Eastern AM basin; cultures have never been obtained and no molecular data are available. The forms of T. cecili resemble those we previously described in lung imprints from Ca. yacare [21], but we cannot attest whether these forms correspond to one or more trypanosome species. We demonstrated that Ca. crocodilus captured in the AM, AT and OR basins can harbour at least three genetically different trypanosomes. Because the holotype of T. cecili is lost, no trypanosome could be elected as a neotype for T. cecili. Therefore, the trypanosomes herein analysed must be described as new species.

Discussion
In this study, we assessed the genetic diversity and phylogenetic relationships of trypanosomes from South American alligatorids and African crocodilids, and compared phylo-and biogeographical patterns of these parasites with those of their host species to uncover possible events that could help to hypothesize evolutionary scenarios. The phylogenetic analyses enabled the description of T. ralphi n. sp and T. terena n. sp from Brazilian alligatorids. Data on culture behaviour, morphology and ultrastructural features complemented the species description.
The crocodilian trypanosomes from Africa and South America clustered together forming the strongly supported Crocodilian clade, which comprised three subclades led respectively by T. terena, T. grayi and T. ralphi. Unexpectedly, a trypanosome from the African O. tetraspis clustered tightly within the South American T. terena subclade instead of the T. grayi subclade exclusive of African trypanosomes. Also unpredictably, the T. ralphi subclade, which comprises exclusively South American isolates, was more closely related to the African T. grayi subclade than to the South American T. terena subclade.
The close genetic relationships between South American and African crocodilian trypanosomes is incompatible with the Cretaceous (~90 mya) split of Crocodylidae and Alligatoridae [28,30,32]. The results are more consistent with the paleontological and phylogenetic evidence of Crocodylus marine circumtropical dispersal at the Miocene/Pliocene boundary (~4-5 mya) [31][32][33]. Therefore, we hypothesised that marine dispersion played an important role in determining the close relationships between South American and African crocodilian trypanosomes. This scenario favours oceanic dispersion as the determinant of the disjunct distribution of crocodilian trypanosomes, that is, the discontinuous distribution of closely related trypanosomes of hosts separated by large geographical distances [40]. Disjunct distribution can also be demonstrated for other trypanosomes such as the homogeneous T. lewisi and T. theileri, which are distributed throughout the word, probably by the anthropogenic dispersal of their respective domestic rat and cattle hosts [11,16,18,39,41,42].
Our findings are in accordance with paleontological and phylogenetic evidence that Crocodylus species colonized the Neotropics and Afrotropics in the late Miocene (~5 mya) through transoceanic circumtropical dispersion. The dispersal of Crocodylus from Australasia proceeded to the Neotropics and then to Africa or inversely to Africa and then to the Neotropics [31][32][33]. Whatever the route, there was a profusion of alligatorids at the South American hydrographic basins when the ancestors of Crocodylus reached the Neotropics [33,35,36,43]. At that time, Amazonia was part of a much larger wetland extending from the OR to the PP basins without barriers separating the large and diverse crocodilian populations. The South American basins reached its present configuration at the Pliocene (~3 mya), following successive disconnections and reconnections due to marine and river drainage changes accompanying the Andean uplift. The present day basins are still interconnected although at a much smaller scale [43][44][45][46]. Resident alligatorids and the new arriving crocodilids shared wetlands and lakes where successive episodes of trypanosome host switches, apparently restricted to phylogenetically related hosts, may have occurred [14,[16][17][18]47,48]. Host switching mediated by ecological fitting appear to be commonly used for trypanosome jumping among hosts that share ecological niches and can, therefore, serve as blood source for the same hematophagous vectors [12][13][14][15][16]19,[47][48][49][50]. However, this part of the biogeographical history of crocodilian trypanosomes requires data about their transmission in South America. Tsetse flies, which cyclically transmit T. grayi, are widespread in Sub-Saharan Africa, and also occur in a few spots of the Arabian Peninsula. However, there is fossil evidence indicating tsetse dispersal into the Nearctic at the Miocene [51]. The virtual absence of tsetse flies in the location at Guinea Bissau where the trypanosome from O. tetrapsis was obtained suggests the existence of alternative vectors for crocodilian trypanosomes in Africa. In South America, different haematophagous vectors could transmit the crocodilian trypanosomes either cyclically or mechanically. We are currently searching for possible vectors among leeches and insects aiming to provide new insights into the evolutionary history of crocodilian trypanosomes.
Comparative phylogeography using parasites as biological tags has been valuable to track vertebrate hosts and vectors in time and space [1][2][3][4][5]18,39,[52][53][54]. The finding that all crocodilian trypanosomes, whatever their vectors, clustered together forming a clade exclusively of crocodilian trypanosomes, regardless of host species and geographic origin, indicated that the evolution of these trypanosomes was strongly linked to that of their vertebrate hosts. Our phylogeographical analysis can contribute to the understanding of the worldwide dispersal of crocodilians, similarly to studies on bat trypanosomes that indicated recent movements of bats that could not be inferred from available paleontological and phylogenetic data [11,39,41]. Furthers studies including more comprehensive sampling of trypanosomes from crocodilians, in Africa, Asia and the Americas, and their vectors are required for an improved hypothesis about the origin, dispersal route and evolutionary relationships of crocodilian trypanosomes.

Conclusions
Phylogenetic and paleontological data on crocodilians and phylogenetic and biogeographical data on their trypanosomes are consistent with past and present biogeographical scenarios playing important roles in the genetic diversity, and phylogenetic relationships of crocodilian trypanosomes. In the most likely scenario, the evolutionary history of crocodilian trypanosomes was shaped in response to the ancient geographical isolation between alligatorids and crocodilids, followed by the transcontinental voyage of Crocodylus at the Miocene allowing the contact between South American alligatorids and crocodilids. This contact probably enabled successive host switches of trypanosomes between resident alligatorids and migrating crocodiles throughout the configuration of the present day South American hydrographic basins. Development (IDSM). We thank Cristina Schwartz for the coordination of the work in Guinea Bissau. We also thank Marcio C. Valentin from the Laboratory of Electron Microscopy, Institute of Biosciences, USP, and Carlos E. Jared and Marta M. Antoniazzi from the Institute Butantan, São Paulo, SP, Brazil, for their kindly permission to use their electron microscopic facilities. This work was supported by CNPq (PROAFRICA and PROSUL) and CAPES (Programa Nacional de Incentivo à Pesquisa em Parasitologia Básica). LBV was postdoctoral fellow sponsored by CNPq-MCTI (PROTAX -National Program of Taxonomy) and CAPES (PNPD). BRF is recipient of a scholarship from CNPq (PROTAX).