Host-Switching Events in Litomosoides Chandler, 1931 (Filarioidea: Onchocercidae) are Not Rampant but Clade Dependent

Litomosoides Chandler, 1931 (Filarioidea: Onchocercidae), currently includes 42 species of vector-borne filarioids associated with mammals of the orders Rodentia, Chiroptera, and Didelphimorphia. Their geographic distribution ranges from the coastal plains of the Gulf of Mexico to the northern half of Argentina and Chile (Forrester and Kinsella, 1973; Notarnicola, 2004; Notarnicola and Navone, 2011; Landaeta-Aqueveque et al., 2014); most of the known species occur in South America, with 5 present north of the Isthmus of Panama. Among the latter group of species, Litomosoides sigmodontis Chandler, 1931, infects cotton rats, Sigmodon hispidus Say and Ord, and Florida mice, Podomys floridanus (Chapman) in localities of the plains of the Gulf of Mexico (Kinsella, 1991). Because of their ability to reproduce in a wide variety of laboratory mammals (Chandler, 1931; Hawking and Burroughs, 1946; Pringle and King, 1968; Siddiqui and Kershaw, 1976; Cárdenas et al., 2010) and develop in several organs of their mite vector, Ornithonyssus bacoti (Hirst) (Williams and Brown, 1945; Hawking and Burroughs, 1946; Williams, 1948), this parasite is widely used as a model in the investigations of the molecular mechanisms of filarioid immune evasion, host–parasite interactions, anthelmintic activity of new drugs, and the effect of bacterial symbionts on transmission success. Although the natural arthropod vectors in the wild for most species of Litomosoides remain unknown, up to the present time, mites have been documented as experimental vectors for only 5 species (Guerrero et al., 2006; Lefoulon et al., 2015).

In stark contrast with the importance of these filarioids as a model to understand basic aspects of parasitism and filarial disease in humans and animals, knowledge of their phylogeny and the study of their evolution has received only sporadic attention (Bain et al., 1989; Bain and Philipp, 1991; Brant and Gardner, 2000; Guerrero et al., 2006; Notarnicola et al., 2010a, 2010b, 2012). Both exhaustive taxon sampling and detailed accounts of the distribution of the parasites in their hosts are important because most tests of events of cospeciation require well-resolved phylogenies and an estimation of the level of host specificity or the host spectrum of the parasites (Light and Hafner, 2008; de Vienne et al., 2013). However, exhaustive surveys of general host–parasite associations are difficult to achieve in a group of parasites with such an extensive distribution. Most of the species of Litomosoides are known only from the original description. Commonly, these old descriptions and combined ancillary data lack any context that may help to understand both the collecting effort and the number of taxa surveyed that will allow current-day researchers to infer any level of host-specificity of the filarioids or the distribution in mammals with sympatric or even syntopic geographic ranges. The archived material backing up those descriptions is or was either degraded beyond the ability of current technology to retrieve molecular data; in some instances, the specimens were not collected with accurate host–parasite vouchering and proper storage of both kinds of organisms (Galbreath et al., 2019). Further, ample taxon sampling is also difficult to complete since very few active biological surveys are being carried out, and any data resulting from these surveys are not easily available. To attempt to circumvent these problems, researchers are turning to the use of statistical approaches that calculate the posterior probability of ancestral distributions and ancestral associations based on phylogenetic patterns and known (current day) parasite associations with hosts (Jiménez et al., 2012).

The presence of these parasites in a broad range of mammalian Neotropical lineages was used to suggest that changes in association are formed through instances of host-switching in which sister species of parasites occur in disparate and phylogenetically unrelated hosts (Bain and Philipp, 1991; Brant and Gardner, 2000). Furthermore, several of the definitive mammal-associated-lineages were involved in the series of faunal exchanges that shaped the Great American Biotic Interchange, GABI (Webb, 1978; Simpson, 1980), which included cricetid rodents, marsupials, and mormoopid bats. As the result of the evaluation of the distribution of species through geographic and definitive hosts, Bain and Philipp (1991) proposed that Litomosoides originated in South American bats, with subsequent switches to other mammalian host lineages including cricetid rodents. This proposal of a single switch to rodents from bats was not supported in a subsequent analysis (Brant and Gardner, 2000), in which rampant host-switching of definitive hosts was determined to be the main force shaping the associations among members of Litomosoides and their mammalian lineages.

However, the phylogenetic reconstruction of species in Litomosoides and the relationships with their putative sister group and other closely related groups are yet to be well-resolved. At the genus level, the poor resolution of the phylogeny is the result of the selection of 22 morphological characters with high levels of homoplasy (Brant and Gardner, 2000). Further, the results of this reconstruction suggest that Litomosoides is not monophyletic and that the use of some characters considered to classify species groups may not have systematic value in classification. Among these characters, the configuration of the spicules is traditionally used to sort species of Litomosoides in species groups; to this effect, the configuration of spicules depends on the proportion among left and right spicules, the proportion of blade and handle in the left spicule, and the ornamentation on the distal part of the right spicule. Traditionally, the species in Litomosoides are grouped in the “carinii” or “sigmodontis” groups (Bain et al., 1989; Notarnicola et al., 2000). Both groups include species parasitizing bats, cricetid rodents, and marsupials. However, in the last 2 decades, several new species were described, and others were re-described or synonymized (Guerrero et al., 2002; Bain et al., 2003; Notarnicola et al., 2010a, 2010b, 2012; Notarnicola and de la Sancha, 2015; Oviedo et al., 2016); thus, some of the Litomosoides species were re-assigned to either one or another group, and, as a result, some of them do not fit with the original definition of the “carinii” or “sigmodontis” groups. Examination of specimens deposited in helminthological collections as well as material obtained from our collecting efforts suggest there may be a third type of spicule, which may be typified by the structure present in Litomosoides hamletti Sandground, 1934.

Interestingly, species of Litomosa Yorke and Maplestone, 1926 (a genus that includes filarioids of bats from the Ethiopian, Palearctic, and Australian regions), were suggested as the potential sister group of species in the genus Litomosoides (see Junker et al., 2009). More recently, a study of the phylogenetic relationships of all available filarioid nematodes for which DNA data could be easily obtained showed that species in the genus Litomosoides share a common ancestor with a group that includes Cruorifilaria Eberhard et al., 1976; Yatesia Bain et al., 1983; and Cercopithifilaria Eberhard, 1980; all of the species that have been studied well are known to use acarine vectors for transmission, and it is expected that the rest of the species in this clade will also show the same life-cycle characteristics.

In the current paper, we use homologous sequences of the mitochondrial genome from available specimens and sequences of species of Litomosoides, Litomosa, and other species of the Onchocercidae. Currently, the phylogenetic relationships of species of Litomosoides are not well-resolved, and the previous phylogeny of the group based on morphological characteristics shows several unresolved polytomies (Brant and Gardner, 2000) that when interpreted in a phylogenetic context show several independent (therefore rampant) events of host-switching from hystricognath rodents to cricetids and then to bats. This proposed evolutionary scenario is fundamentally different from that proposed by Bain and Philipp (1991), who suggested an early origin in Chiroptera with a subsequent transfer to rodents of the families Cricetidae, Geomyidae, and Ctenomyidae.

We herein resolve these relationships by using additional characters to reconstruct the phylogeny of the group. Our dataset includes specimens collected from extensive biodiversity surveys that have properly documented and vouchered parasites and hosts deposited in museum collections. Further, we propose to use this phylogeny to estimate the ancestor–descendant associations among the ancestor of Litomosoides and bats and the putative switch to cricetid rodents. This phylogeny allows us to address 4 main objectives in this paper: (a) test the monophyly of Litomosoides and simultaneously test the potential usefulness of their spicules in classification; (b) reconstruct the putative origin of the genus Litomosoides in the Neotropical or Nearctic region, (c) estimate the ancestral association between Litomosoides and bats; and (d) test a hypothesis of a single origin of the species of Litomosoides in rodents of the family Cricetidae.

From 1984 to 2000, the American Museum of Natural History (AMNH), the Museum of Southwestern Biology (MSB), the Harold W. Manter Laboratory of Parasitology (HWML), and the Bolivian National Museum of Natural History in La Paz mounted joint collecting expeditions throughout Bolivia to survey and inventory sylvatic mammals and their parasites. Additional work was conducted in Argentina as part of the research program of JN and complemented with material from the Argentina Project directed by Janet Braun and Michael Mares (University of Oklahoma). During field-work Sherman live-traps and Museum Special snap traps were used to collect small mammals. To collect bats, mist nets were deployed at the same localities. In the case of filarioids presently discussed, nematodes found were removed from either the thoracic or abdominal cavities, washed in saline, and either preserved in 10% formalin and/or 95% ethanol, others were placed in cryotubes, flash-frozen in liquid nitrogen, and stored at −85 C (Gardner and Jiménez-Ruiz, 2009). In the laboratory, worms were measured for identification, and some of them were cut into fragments, preserving the anterior and posterior extremities for morphological study while the remaining pieces were used for extraction of genomic DNA. Some complete specimens and some fragments that included diagnostic characters were cleared with lactophenol for morphological study, and these specimens were subsequently deposited in the Parasite Collection of the Harold W. Manter Laboratory of Parasitology, University of Nebraska State Museum (Lincoln, Nebraska), and the Helminthological Collection of the Museo de La Plata (La Plata, Argentina). A list of hosts, their localities, and the fate of the voucher specimens of Litomosoides and their mammalian hosts is provided in Table I.

Genomic DNA was extracted using Chelex beads (BioRad, Inc. Hercules, California). A 520-bp fragment of the mitochondrial 12S rDNA and a 640-bp fragment of the cytochrome oxidase 1 (COI) were amplified using primers and thermal profile described elsewhere (Casiraghi et al., 2001, 2004). Reactions used approximately 100 ng of genomic DNA in volumes of 20 μl using the PCR Core Kit, following manufacturer recommendations (Qiagen Inc., Valencia, California). Successful amplicons were purified using ExoSAP-IT (GE Healthcare, Cleveland, Ohio) following the manufacturer's recommendations. Purified products were processed with BigDye 3.2 (Applied Biosystems Inc., Foster City, California) and direct sequenced in a Base Station 51 DNA Fragment Analyzer (MJ Research, Inc., Waltham, Massachusetts).

The resulting sequences were aligned with homologous sequences linked to voucher specimens (Casiraghi et al., 2001, 2004; Junker et al., 2009; Lefoulon et al., 2015). A list of these specimens and their sequences uploaded to GenBank is available in Table I; the resulting matrix includes 47 taxa. We used Akaike information criterion to select the best model of evolution for this matrix, performing this analysis using jModeltest V. 2.1.6 (Posada, 2008) as implemented in CIPRES (Miller et al., 2010). Phylogenetic signal was analyzed using PAUP* and RAXML using parsimony and Maximum Likelihood (ML) as optimality criteria (Swofford, 2003; Stamatakis, 2014). The most parsimonious trees were found through a heuristic search, and a heuristic bootstrap of 1,000 replicates was performed to estimate clade support. For the phylogenetic reconstruction using ML, each gene was considered a partition with its own model of evolution, and a parametric bootstrap was implemented to assess branch support. MrBayes v3.2.7 (Ronquist et al., 2012) was used to infer the posterior probabilities of the nodes; this analysis was run for 10 million generations with a burn-in of 25%. Results were evaluated for chain convergence by difference in standard deviation. The resulting trees were uploaded to a universal data repository (https://opensiuc.lib.siu.edu/zool_data/16/).

Historical associations among parasites and hosts were reconstructed using the Biogeography with Bayesian and Likelihood Evolutionary Scripts (BioGeoBEARS) approach as implemented in the program Reconstruct Ancestral States in Phylogenies (RASP) Version 4.2 (Matzke, 2013; Yu et al., 2020). The approach was implemented in a universe of 21,001 phylogenetic trees to optimize the vicariant or cospeciation events at the cost of reducing the number of possible events of dispersal vicariance (DIVA), dispersal extinction and cladogenesis (DEC), host-switching, or extinction. In particular, DEC offers an estimate of the likelihood of ancestral states among possible inheritance states (ranges, hosts) at a given cladogenetic or speciation event or node. These competing likelihoods are calculated using the instantaneous transition rates between possible discrete states (ranges, hosts) along a branch, repeated in each tree present in a universe of trees. In addition, a Bayesian Reversible-jump Markov Chain Monte Carlo simulation as implemented in BayesTraits 3.0.2 (Pagel et al., 2004) was used to reconstruct the hosts and putative areas of origin and dispersal for relevant nodes of Litomosoides + Litomosa.

To reconstruct the possible origin, dispersal, and affinities of the parasites with their host lineages, each of the terminal taxa were scored according to the family to which these mammals are assigned. For example, in our analysis, Litomosoides brasiliensis Lins de Almeida, 1936, was recorded from 3 species of bats in the Phyllostomidae; thus, that taxon was used as a host lineage. The definition of this scoring system is available in Table II.

The phylogenetic tree obtained in the calculation of the branch support via MrBayes was pruned to include only Litomosa chiropterorum Ortlepp, 1932; Litomosa westi (Gardner and Schmidt, 1986); and Litomosoides spp. The 21,001 trees that remained after the burn-in from the Bayesian estimation of clade credibility were combined and trimmed to include only the relevant taxa; these trees were used as the basis for the analyses and are available from (https://opensiuc.lib.siu.edu/zool_data/16/).

In the case of the approach BioGeoBEARS, 1,000 random trees were used to optimize the events. All tests were run with no constraints relative to the potential cladogenesis, dispersal, or extinction of the groups. The same number of replicates and the same universe of trees were used to compute the likelihood values for statistical DIVA and statistical DEC running a thousand replicates. The weighted Akaike Information Criterion was used as a criterion to select the model that best describes the data (Yu et al., 2020).

We used this universe of 21,001 trees during the Bayesian inference of posterior probability of branches and fossilized the internal node using the host lineage described in Table II. BayesMultistate allowed the free host change among the mammalian lineages listed in Table II. In total, 3 million iterations were performed for each analysis, with the first 10% being discarded as burn-in with sampling every thousandth generation. In total, 100 stepping stones were enforced for 10,000 iterations. The log-normal marginal likelihood values were converted into log Bayes Factors to test competing hypotheses (Pagel et al., 2004). Each analysis was performed 3 times to increase the effective estimation of the marginal likelihood.

To test for the possible dispersal of the parasites and to estimate their putative area of origin, each terminal was scored according to their collection continent (Table II). The ancestral geographic distribution of the filarioids was then estimated using the methods described above. In this case, we imposed a restriction on the transition between Africa and North America and vice-versa.

In the classical systematic classification of these nematodes, 2 species groups were proposed (Bain et al., 1989). One is called the “carinii” group and is recognized to include only species in the genus Litomosoides that possess a left spicule with a handle longer than the blade (Fig. 1). In addition in the “carinii” group, the right spicule has a well-sclerotized dorsal heel and terminal cap (Fig. 2) (Bain et al., 1989; Notarnicola et al., 2000). The other is called the “sigmodontis” group and includes species of Litomosoides in which males have a left spicule with handle (calomus and manubrium) shorter than or equal in length to the blade or lamina. The blade in this group is divided into anterior membranous folded alae terminating distally in a thin filament (Fig. 3); the right spicule features a heel that is not heavily sclerotized (Fig. 4) (Bain et al., 1989; Notarnicola et al., 2000).

A third type of spicule morphology has also been defined and is termed the “hamletti” group. In these nematodes, the left spicule has a handle (calomus and manubrium) longer than the blade (or lamina). The lamina is composed of an anterior part with 2 well-sclerotized rods, terminating in a thin filament (Fig. 5); the right spicule features a heel not heavily sclerotized with a rounded extremity sporting a dorsal notch giving it a hook-like appearance (Fig. 6) (Esslinger, 1973; Oviedo et al., 2016). The “hamletti” group of species include L. hamletti, Litomosoides chandleri Esslinger, 1973, and Litomosoides yutajensis Guerrero, Martin, and Bain, 2003, studied herein.

To estimate the putative origin of the 3 groups of species, the terminal species on the phylogeny were scored as either “carinii,” “hamletti,” or “sigmodontis” according to the characterization made in the original descriptions and our observations and determination of the morphological characteristics of the spicules in the specimens examined here. Since the spicule type in species of Litomosa does not correspond to any of these groups, the spicule was scored as “other morphology” (Table II). The ancestral state was reconstructed allowing instantaneous transitions along the branches using the methods described above for Bayes Traits with no restriction in the transition among the 4 character states.

A total of 12 equally parsimonious trees (https://opensiuc.lib.siu.edu/zool_data/16/) resulted from the analysis of 2 mitochondrial markers using parsimony as the optimality criterion. From 1,078 characters, a total of 459 were phylogenetically informative; the length of the tree is 2,141 steps, the consistency index (CI) = 0.426, and the homoplasy index (HI) = 0.575. A clade including L. westi and L. chiropterorum appears to be sister to a clade including all species of Litomosoides; in turn, species of Cercopithifilaria appear as the sister group for Litomosa and Litomosoides. The bootstrap consensus for the other genera of Onchocercidae Clade ONC4 including Achanthocheilonema, Monanema, Crurofilaria, Yatesia, Cercopithifilaria (lower support values on Fig. 8) is weak.

The topology resulting from the analysis using maximum likelihood as the optimality criterion, as well as bootstrap branch support, is shown in Figure 8; in this topology L. westi is nested within Litomosoides, and this clade appears as the sister group for L. chiropterorum, with branch support of 85%. In turn, Yatesia hydrochaerus (Yates, 1980) Bain, Baker, and Chabaud, 1982, appears to be the sister group for Litomosa + Litomosoides. Support is less than 50% for the branches defining the relationships among species in Cercopitifilaria, Crurofilaria, Monanema, and Yatesia.

The tree resulting from the Bayesian estimation of clade credibility is shown in Figure 7. Litomosoides appears as a monophyletic group, yet Litomosa is not. The posterior probability for the branch including Litomosoides and L. westi is 0.84 (Fig. 7 and Node 54 in Fig. 9). The branch including all Litomosoides has a posterior probability of 0.73 (Fig. 7 and Node 53 in Fig. 9). Litomosoides brasiliensis and then Litomosoides solarii Guerrero, Martin, Gardner, and Bain, 2002, branch close to the Litomosoides node, from these, only the node including specimens of L. brasiliensis features the maximum posterior probability and bootstrap support (Fig. 8, Node 37, Fig. 9). The rest of the species of Litomosoides are included in Node 52, with a posterior probability of 1 and branch support of 94% (Figs. 7, 8); 2 internal clades contain the bat-dwelling species L. hamletti (Node 50, Figs. 8, 9), and L. chandleri plus L. yutajensis (Node 47, Fig. 9). Our trees show that all the species of filarioids (Litomosoides) parasitic in the Cricetidae share a common ancestor (Node 45, Figs. 7–9) and are allocated to 4 clades, including: Litomosoides bonaerensis Notarnicola, Bain, and Navone, 2000, and Litomosoides scotti Forrester and Kinsella, 1973 (Node 44, Fig. 9); Litomosoides odilae Notarnicola and Navone, 2002 (Node 42, Fig. 9); an (as yet) unnamed species of Litomosoides; and an unresolved clade that contains L. sigmodontis; Litomosoides galizai Bain, Petit, and Diagne, 1989; Litomosoides esslingeri Bain, Petit, and Diagne, 1989; and Litomosoides taylori Guerrero and Bain, 2011 (Node 40, Fig. 9).

Our analysis shows that species of Litomosoides had an early origin in bats, probably of the family Phyllostomidae (Node 53, Fig. 9A; Table III). The solution identifies 2 switches, one to cricetid rodents (Node 45, Fig. 9A; Table III) and the other one to mormoopid bats (Node 46, Fig. 9A; Table III). These events of host-switching are supported by the estimation of fossil states at 2 nodes: Node 45 from phyllostomid bats to cricetid rodents (Bayes Factors = 7.487) (Table III); and Node 46, from phyllostomid to mormoopid bats (Bayes Factors = 11.05).

The solution for the reconstruction of the area of origin for species comprising the genus Litomosoides suggests a South American origin for these nematodes (Node 53, Fig. 9B; Table III). The reconstruction for Node 40, Node 44, and Node 46 suggests taxon pulse cladogenesis and simultaneous dispersal into North America. The South American origin of Litomosoides is not supported by Bayes Factors values (Node 53 in Table III). Finally, the taxon pulse dispersal into North America is supported in 2 clades, Node 38 (Fig. 9B), which includes L. sigmodontis (Bayes Factor = 15.679) and Node 44 (Fig. 9B) which includes L. scotti (Bayes Factor = 6.084).

The reconstruction of the ancestral state in the fossilized Node 53 results in equally likely states as spicule types “carinii” or “hamletti.” Both types feature similar Bayes Factors of 6.811 and 6.673, respectively. A similar pattern was observed for Node 52, which excludes L. brasiliensis, and for Node 51. For the latter, the Bayes Factors for spicule types “carinii” and “hamletti” are 7.765 and 7.102, respectively; whereas they are 7.429 and 7.939 for types “carinii” and “hamletti” in Node 52. The “hamletti' type apparently originated twice, on Node 47 including the common ancestor of L. chandleri and L. yutajensis (Bayes Factors = 5.423), and again in Node 50 including the common ancestor of L. hamletti (Bayes Factors = 13.78). The “carinii” type characterizes the spicules present in L. brasiliensis, which were reconstructed as the ancestral state for Node 37 with Bayes Factors of 21.369. Finally, the transition to the “sigmodontis” type occurs in Node 41 with a Bayes Factor = 14.263.

We present the first phylogenetic hypothesis for species of Litomosoides using mitochondrial genes. This phylogeny suggests that Litomosa is firmly in clade Onchocercidae ONC 4 (Lefoulon et al., 2015). Litomosoides appears to be sister to Litomosa, yet the latter appears to be polyphyletic. Both Litomosoides + Litomosa are part of a non-resolved clade that includes species of Acanthocheilonema, Cercopithifilaria, Crurifilaria, Monanema, and Yatesia. Further, the topology suggests that species of Litomosoides originated in South American bats, that species of Litomosoides infecting rodents of the family Cricetidae form a clade, and that the distribution of these species in North America is explained by taxon pulse diversification and dispersal. The identification of 2 host-switching events involving mormoopid bats and cricetid rodents hints that the faunal changes involved in the GABI may have played an important role in the diversification of these parasites.

The tree shown in Figure 7 suggests that Litomosoides is a monophyletic group. This finding is consistent with the proposal of the elongated buccal cavity as a synapomorphy for the group. The results also suggest that Litomosa may be the sister group for Litomosoides, yet Litomosa is polyphyletic if L. westi is retained in this genus (Figs. 7, 8). Conversely, Figure 8 suggests that Litomosoides is paraphyletic precisely because of the placement of L. westi relative to L. solarii and L. hamletti. These relationships contradict the placement of L. westi as an outgroup of Litomosoides as suggested in an analysis of morphological characters (Brant and Gardner, 2000). Rather, the relationships shown in Figure 8 would reinforce the notion that L. westi should be reinstated in the genus Litomosoides, as originally proposed based on the characters of the buccal capsule and left spicule (Gardner and Schmidt, 1986). This taxonomic consideration cannot be solved by our present analysis because we lack relevant sampling of other species of Litomosa from around the globe. Although Node 59, which includes L. westi, L. chiropterorum, and Litomosoides spp., is present in the majority of the trees and enjoys a relatively robust posterior probability, the clades of L. westi and Litomosoides spp. are not well-supported, as suggested by the bootstrap values of the parsimony and ML reconstructions (Fig. 8). This lack of bootstrap support makes it appear as if L. westi (Node 32), L. brasiliensis (Node 37), and the species included in Node 51 form a polytomy. This phenomenon appears to be consistent with the topological effects of long-branch attraction described elsewhere (Felsenstein, 1978; Anderson and Swofford, 2004). The clade containing L. westi may be pulled toward L. brasiliensis and L. solarii, and these are clades that lack significant bootstrap support in their inclusion with the other species in the genus. We anticipate that the resolution of the relationships between Litomosa and Litomosoides will be achieved by increasing a wider genome coverage and including additional species of both phyllostomid dwelling Litomosoides and species phylogenetically close to L. westi.

The results suggest that the common ancestor of Litomosa + Litomosoides predated the splitting of Gondwanaland into the continental plates of the southern hemisphere. This is supported by values for Bayes Factors recovered by the optimization of both continental origins in the reconstruction of the nodes indicated by fossils (Fig. 9B; Table III). This same pattern also appears in the reconstruction of the ancestral host for this node, which appears to be an ancestor of bats in the families Miniopteridae and Phyllostomidae. Although our results appear to be consistent with the breakup of Gondwanaland, they should be interpreted in this light with caution. This is because the phylogenetic tree that we used lacks the properties to test this hypothesis via time calibration and the addition of more species of Litomosa in the analysis is necessary to test competing hypotheses. Relevant species include representatives of Litomosa present in bats occurring in the Australian and Oriental regions.

The South American origin of Litomosoides (Node 53, Fig. 9B), produces a strong signal detected through the DEC analysis, yet it does not enjoy strong positive evidence according to the comparison of Bayes Factors. The latter may be a direct result of the lack of resolution of the phylogenetic trees employed for the analysis. Nevertheless, the reconstructed ancestral host associate lineage for this node is the Phyllostomidae, which shows very strong support in both analyses employed (Fig. 9A, B; Table III).

Most of the nodes including species of Litomosoides appear to be associated with phyllostomid bats in South America; this is the case of L. brasiliensis (Node 37), L. hamletti (Node 50), and L. solarii. Yet, Nodes 41, 45, and 46 feature changes in their host associations, showing evidence of host-switching to cricetid rodents and mormoopid bats. In the case of L. yutajensis (Node 46), the reconstruction of ancestral geographic distribution suggests a geographic expansion from South America into North America without diversification; further, this solution indicates that the ancestral host was a bat of the family Mormoopidae, in a host-switching event supported by very strong evidence (Bayes Factors = 11.05; Table III). This reconstruction is congruent with the tectonic/geographic changes that brought North and South America closer together and enabled the mixing of faunas. Interestingly, bats of the family Mormoopidae may have originated in the northern Neotropics, in an area outside the geographic realm of South America (Dávalos, 2006). The Mormoopidae includes 8 extant species that occur across North and South America with a hypothesized center of origin in the northern Neotropics, closely associated with the transition zone between these continents (Dávalos, 2006). The diversification of this group and their subsequent dispersal and use of the subtropical lands in South America may have contributed to host-switching via cladogenesis or taxon pulse and subsequent exposure of these bats to the vectors of phyllostomid dwelling Litomosoides. Furthermore, this reconstruction is congruent with the hypothesized relatively recent dispersal of the mormoopid clade into South America, dated as recently as 6 million years ago (MYA) (Pavan and Marroig, 2017). Thus, species of Litomosoides would have been exclusively parasites of fruit-eating phyllostomid bats until invading species representing insectivorous bats colonized the area via dispersal. Thus far, L. yutajensis has been found infecting Pteronotus parnelli in the Amazon forest of Venezuela and Mexican dry forest (Guerrero et al., 2003; 2006; present study). It is still unclear why other native species of insectivorous bats do not harbor infections with these or related species of filarioid nematodes.

The second event of host-switching and simultaneous geographic dispersal involves the species of Litomosoides that infect rodents of the family Cricetidae and descend from the hypothetical ancestor represented by Node 45. For this, there is strong evidence of a South American origin for Node 45 that corresponds with a host-switch from phyllostomid bats into cricetid rodents (Fig. 9A, B; Table III). The results for Node 45 also indicate an episode of geographic expansion that resulted in 2 branches (Nodes 40 and 44) each with representatives in both North and South America. This is arguably the result of a taxon pulse evolutionary event that simultaneously enabled ecological fitting as cricetid rodents dispersed into South America and were exposed to both autochthonous blood-sucking arthropods (possible vectors of pathogens) and autochthonous mammals and their parasites (Jiménez et al., 2017). The dispersal of these rodents is estimated to have occurred during the Neogene, about 12 MYA (Parada et al., 2013), with subsequent patterns of diversification following the spread of these rodents across the previously isolated landmass (Maestri et al., 2018). With this subsequent diversification of cricetid rodents in South America, a small number of lineages dispersed back into North America (Parada et al., 2013; Maestri et al., 2018). Perhaps these ancestral cricetids carried ancestral lineages of parasites that are now common in tropical and subtropical areas of Central and North America, including ancestors of L. sigmodontis and L. scotti; both of these species occur across the coastal plains of the Gulf of Mexico. Considering a host-switching event simultaneous with the early arrival and diversification of cricetid rodents into South America, this event could have occurred no earlier than 12 MYA, which was the earliest calculated time of arrival of the putative hosts as suggested by calibrated phylogenies and the fossil record (Parada et al., 2013). Although the evolutionary event may have occurred before the formation of the Panamanian Isthmus, it could pinpoint an event of host-switching involving mammals typically associated with GABI (Jiménez et al., 2017). In this case, the diversification and southward dispersal of the invading cricetid rodents would have been exposed to the vectors carrying microfilaria of Litomosoides. A more exhaustive geographic field-based sampling and genome-wide analysis of characters may help to determine these changes in more detail.

The taxon identified as Litomosoides sp. included in Node 41 occurs in South America, yet it was extracted from the cavity of a murid that was an unidentified species of Rattus. The Muridae includes species of synanthropic rodents that were recently introduced into the New World, and up to the current time, 3 records of species of Litomosoides have been documented from murid rodents, including L. sigmodontis in Venezuela; Litomosoides pardinasi Notarnicola and Navone, 2011, in Chile; and the specimens of Litomosoides sp. from Argentina herein reported (Vogel and Gabaldon, 1932; Landaeta-Aqueveque et al., 2014). It is possible that these 3 species of filarioid nematodes are cycling in endemic cricetid rodents. Thus, the exposure of an individual rat to the vector may have facilitated infection. The presence of L. pardinasi in conspecific individuals in wild and synanthropic murid rodents supports this pattern. We speculate that the presence of this parasite in these rodents is the result of recent-time ecological fitting (Janzen, 1985), a phenomenon in which the parasites become established in a group of compatible hosts, even when their association was not shaped by a shared evolutionary history.

The topology of the tree shows the species categorized as “carinii” forming a paraphyletic conglomerate. Species in this category or group include the cricetid-dwelling L. bonaerensis, L. odilae, and L. scotti, and the phyllostomid dwelling L. brasiliensis. Species included in the “hamletti” group also appear to be a paraphyletic assemblage, since it consists of the phyllostomid dwelling L. chandleri and L. hamletti plus the mormoopid dwelling L. yutajensis. In turn, the “sigmodontis” group is strongly supported by Bayes Factors (8.725) at Node 41; this clade includes Litomosoides sp. (a filarioid from Rattus in Argentina), L. esslingeri, L. sigmodontis, and L. galizai.

Our analysis shows clearly that the only node with a strong signal for both the major host-switching event to cricetid rodents, the origin of a novel morphological structure of the spicules, and a putative event of cladogenesis is Node 41. These 3 phenomena concentrated in 1 clade support the conclusions offered by Brant and Gardner (2000), who detected that filarioids featuring this spicule type formed a monophyletic group with a high consistency index (CI = 1). By finding support for this clade using a set of independent characters, we underscore that the “sigmodontis” spicule type is an actual synapomorphy for the cricetid-dwelling species since these species share a common ancestor. As indicated earlier, the inclusion in a phylogenetic analysis of additional species infecting bats may help resolve the relationships among L. brasiliensis, L. hamletti, and L. solarii, and this action followed by the corresponding analyses may help test the monophyly of the “carinii” group.

The underlying biological factors that facilitated these ancient events of host-switching via episodes of ecological fitting described above remain to be discovered. Experimental infections have been carried out to test host specificity of several species in Litomosoides (Hawking and Burroughs, 1946; Guerrero et al., 2003), yet these tests have not included representatives of all species included in the genus. The generalist habits of tropical rat mites, their survival away from the body of the host, and their phoretic abilities make the unwary researcher suspect that the species of Litomosoides may be compatible with these mites. Yet, the distribution and diversity of macronyssid mites are not fully understood, and the phylogenetic relationships among members of Ornithonyssus are yet to be resolved (Dowling and OConnor, 2010; Nieri-Bastos et al., 2011). Furthermore, the origin and distribution of O. bacoti across South America has not been properly established because the name of the tropical rat mite has been used to characterize different species, several of which may be hard to differentiate (Nieri-Bastos et al., 2011). Lareschi et al. (2003, 2007) stated that the presence of O. bacoti parasitizing the cricetid Oxymycterus rufus is associated with the presence of the filarioid Litomosoides oxymycteri Notarnicola, Bain, and Navone, 2000, and another species of acarine are associated positively with L. bonaerensis parasitizing Oligoryzomys spp. (i.e., Laelaps paulistanensis Fonseca and Gigantolaelaps wolffsohni (Oudemans)), as well as the lice Hoplopleura travassosi Werneck, 1932. These reports indicate that additional hematophagous species could be involved in the natural transmission of Litomosoides spp. (Notarnicola, 2004; Oviedo et al., 2016). Furthermore, in these studies, the highest prevalence values were registered in acarian species other than O. bacoti (<70% for G. wolffsohni and Mysolaelaps microspinosus Fonseca). In contrast, in vespertilionid and phyllostomid bats the most frequently recorded ectoparasites included dipterans (Streblidae) and in less frequency, ticks and mites (Autino et al., 1999; Autino and Claps, 2000; Dick et al., 2007). In addition, there is no clear experimental evidence about the vectors transmitting species of Litomosoides that infect both phyllostomid and mormoopid bats. Yunker and Chitwood (1973) found a macronyssid naturally infected with a filarioid juvenile from Artibeus jamaicensis in Venezuela, and available experimental work suggests that the infective stage of the bat-dwelling species of Litomosoides will develop in the tropical rat mite, yet their maturation seems to be suboptimal compared with development of the infective stage for rodent-dwelling species in these vectors (Guerrero et al., 2003). We suspect that although arguably compatible, O. bacoti is not the usual intermediate host for bat-dwelling filarioids.

The present phylogeny was estimated with only 25% of the known diversity of the group, and it does not include species known to infect marsupials, hystricognath, or sciuromorph rodents. Their inclusion would be relevant because estimating the relationships of these parasites and characterizing their genetic diversity may help to test the role of ecological fitting or host-switching in the present-day associations of members of the genus. Both didelphid marsupials and hystricognath rodents collectively include more than 300 species; they diversified extensively through all of South America before the arrival of the cricetid/sigmodontine rodents and sciurids from the north. Yet, only 6 species of Litomosoides are known to occur in these mammals from a handful of localities, a relatively small number since one would expect a greater number of species of Litomosoides in these mammals. Since most of these species feature the “sigmodontis” spicule type, we predict that the association of these 6 species of Litomosoides with marsupials and hystricognath rodents resulted from events of ecological fitting that may have consolidated as host-switching events just recently.

Because of the large number of new species of filarioid and other nematodes that our teams find when we survey endemic mammals, we suspect that only the tip of the iceberg has yet been discovered, and in-depth and concerted studies on the not-so charismatic helminth parasites will yield many more undescribed species. The inclusion of additional species in our hypotheses of phylogenetic relationships is necessary to facilitate a more rigorous reconstruction of the putative host-switching events involving marsupial and hystricognath-dwelling species as well as to resolve the relative position of the Nearctic L. westi and the monophyly of Litomosoides. Relative to the other 3 objectives of our research, our results show that the genus originated in Neotropical bats and that 2 lineages of Litomosoides underwent a geographic expansion, with host-switching events to the Mormoopidae and the Cricetidae with subsequent diversification in the latter.

This study was part of a Fulbright-CONICET (Consejo Nacional de Investigaciones Científicas y Técnicas) scholarship to JN. We express our gratitude to Luis García, Colección Nacional de Helmintos, and Dr. Gabor Racz, Harold W. Manter Laboratory, for facilitating access to material under their care. To Ulyses F. Pardiñas from Mammalian Collection Centro Nacional Patagónico for the identification and deposition of host species into the collection. Mike Kinsella provided valuable material. To Janet K. Braun, Daniel Udrizar Sauthier, M. del Rosario Robles, Lorena Zonta, Ramiro Almagro, Ulyses Pardiñas, Carlos Galliari, Marcela Lareschi, and Graciela Navone for their help during the field collection. Work in the field, laboratory analysis, and writing of this paper were supported by National Science Foundation grants BSR-8612329, BSR-9024816, DEB-9496263, DEB-9631295, DBI-0646356, and DBI-0097019 to SLG; DEB 0103711 to Michael A. Mares and Janet K. Braun; DUE 1564969 to FAJ; and Agencia Nacional de Promoción Científica y Tecnológica 2007 PICT-33816, PICT-33019, UNLP 11N520; 11N627 to JN.