Complex History of Codiversification and Host Switching of a Newfound Soricid-Borne Orthohantavirus in North America

Orthohantaviruses are tightly linked to the ecology and evolutionary history of their mammalian hosts. We hypothesized that in regions with dramatic climate shifts throughout the Quaternary, orthohantavirus diversity and evolution are shaped by dynamic host responses to environmental change through processes such as host isolation, host switching, and reassortment. Jemez Springs virus (JMSV), an orthohantavirus harbored by the dusky shrew (Sorex monticola) and five close relatives distributed widely in western North America, was used to test this hypothesis. Total RNAs, extracted from liver or lung tissue from 164 shrews collected from western North America during 1983–2007, were analyzed for orthohantavirus RNA by reverse transcription polymerase chain reaction (RT-PCR). Phylogenies inferred from the L-, M-, and S-segment sequences of 30 JMSV strains were compared with host mitochondrial cytochrome b. Viral clades largely corresponded to host clades, which were primarily structured by geography and were consistent with hypothesized post-glacial expansion. Despite an overall congruence between host and viral gene phylogenies at deeper scales, phylogenetic signals were recovered that also suggested a complex pattern of host switching and at least one reassortment event in the evolutionary history of JMSV. A fundamental understanding of how orthohantaviruses respond to periods of host population expansion, contraction, and secondary host contact is the key to establishing a framework for both more comprehensive understanding of orthohantavirus evolutionary dynamics and broader insights into host–pathogen systems.


Introduction
Comparative phylogeographic study of parasites and their hosts can provide valuable insights into evolutionary history and processes of diversification for both host and parasite. These analyses also provide a predictive framework for emerging disease under scenarios of changing environmental conditions [1][2][3]. The classic model of strict host-parasite codiversification as the leading process driving parasite evolution has been thrown into question over the past several decades with studies suggesting that host switching is not only more prevalent than originally thought, but that it may be the most common pattern throughout parasite evolutionary history [4,5]. infection in deer mice (Peromyscus maniculatus) [26]. All tissue samples were frozen at -70 • C until tested by RT-PCR.

RNA Extraction, cDNA Synthesis, and RT-PCR Amplification
Total RNA was extracted from 20-50 mg of each tissue using the PureLink Micro-to-Midi total RNA purification kit (Invitrogen, San Diego, CA, USA). cDNA was prepared using the SuperScript III First-Strand Synthesis System (Invitrogen) and oligonucleotide primer (5'-TAGTAGTAGACTCC-3') designed from the conserved 3'-end of the S, M, and L segments of orthohantaviruses.
Gene amplification was carried out in 20-µL reaction mixtures containing 250 µM dNTP, 2 mM MgCl 2 , 1 U of AmpliTaq polymerase (Roche, Basel, Switzerland), and 0.25 µM of oligonucleotide primers, designed from highly conserved regions of previously identified soricid-borne orthohantaviruses. A listing of the oligonucleotide primers used to amplify the S-, M-, and L-genomic segments is provided in Table S1. Initial denaturation was followed by touchdown cycling (two-degree step-down annealing from 48 • C to 38 • C for 40 sec) and elongation at 72 • C for 1 min, then 32 cycles of denaturation at 94 • C for 40 sec, annealing at 42 • C for 40 sec, and elongation at 72 • C for 1 min, in a GeneAmp PCR 9700 thermal cycler (Perkin-Elmer, Waltham, MA, USA). Amplified products were Viruses 2019, 11, 637 4 of 15 separated by electrophoresis on 1.5% agarose gels and purified using the QIAQuick Gel Extraction Kit (Qiagen, Hilden, Germany). DNA was sequenced directly using an ABI Prism 377XL Genetic Analyzer (Applied Biosystems Inc., Foster City, CA, USA).

Sequence Dataset
Virus sequences, either generated in this study or downloaded from GenBank [27] of all currently known strains of JMSV and their respective hosts spanning the range of JMSV, as well as outgroups for rooting of the trees, were studied ( Figure S1A). The final dataset, with outgroups, was composed of partial coding regions for the L (n = 31), M (n = 10), and S (n = 29) segments of orthohantaviruses. In addition, mitochondrial cytochrome b (cyt b) gene sequences, generated for each mammalian host across the range of JMSV variation, were included to independently examine host phylogenetic relationships ( Figure S1B). GenBank accession numbers for all sequences used in this study are in Table S2.

Phylogenetic Analysis
Phylogenetic trees were generated from alignments of each individual genomic segment (S, M, and L) and Ash River virus from Sorex cinereus [14] as an outgroup using Muscle version 8.1.9 [28], as implemented in Geneious 8.1 [29]. To detect instances of intergenic rearrangement and assure we were analyzing homologous genomic regions, these independent alignments were tested for recombination using the Phi [30], NSS [31], and Max χ 2 [32] tests as implemented in PhiPack [33]. Both a maximum likelihood approach using RAxML version 8.2.12 [34], and Bayesian probability implemented in MrBayes version 3.2.5 [35], were employed for tree inference. A general time reversible (GTR) model of nucleotide evolution with gamma-distributed rate heterogeneity and invariable sites (RAxML option GTRGAMMAI) was determined to be the best fit model for this dataset by JModelTest [36]. RAxML generated 1000 bootstrap replicates to determine the best-fit maximum likelihood tree and associated nodal support. MrBayes was run for 10 million generations using the priors from JModelTest by sampling trees every 1000 generations. After the first 25% of trees were discarded as a recommended burn-in, the remaining trees were used to calculate a 50% majority rule consensus tree. Sequence alignments for all segments, as well as tree files in newick format, are available in supplementary materials.

Tanglegrams, Diversity Analyses, Codiversification Tests, and Reconciliation
Tajima's D statistic for both virus and host alignments were computed in R [37] with the package Pegas [38]. Negative values for Tajima's D can be indicative either of population expansion or of a selective sweep [39]. Individual pairwise distances and between group mean distances were calculated in Mega7 [40] with a Kimura 2-parameter model and 500 bootstrap replicates. To test codivergence versus host switching, several metrics that measure the extent of similarity between phylogenetic trees were employed. The nPH85 metric [5] is a normalized version of the Robinson Foulds tree topology distance that incorporates branch lengths and returns a value between 0 and 1, with 0 indicative of strict codiversification between identical tree topologies or 1 indicative of cross species transmission with completely incongruent tree topologies, and was calculated in the R package NELSI [41]. Trip [42] and TripL [43] measure the number of shared triplets between two trees, with TripL incorporating branch lengths and returning the proportion of triplets not shared between two trees, and was calculated in the R package Kaphi [44]. In this metric, 0 indicates identical trees and 1 indicates no shared triplets. Phylogenetic reconciliation was performed for the L and S segment with host cyt b phylogeny in Jane 4.01 [45] with equal costs of 1 assigned to all possible events. Finally, to visualize tree discordance, tanglegrams were generated for each tree comparison in R with the package phytools [46].

Phylogenetic Analysis
Significant values for recombination analyses differed among segment and test used but identified no regions of intragenic rearrangement in any segment analyzed. The JMSV strains showed a pattern of diversification with defined clades that closely mapped to geographic regions similar to that seen in the phylogeographic structure of the hosts, with the exception of the virus recovered from S. palustris, which was previously identified as Fox Creek virus ( [47]; Figure 1). Geographic regions correspond to Northern Continental (NC) and Southern Continental (SC) clades composed of viruses recovered strictly from S. monticola. The NC clade contained JMSV strains from British Columbia and Yukon Territory, as well as Alaska, while the SC clade was found in New Mexico and Colorado. A third clade was recovered from the Pacific coast (PC) which, in comparison to the NC and SC clades, was hosted by several species of the S. vagrans complex, including S. vagrans, S. trowbridgii, S. bairdi, and S. palustris. Phylogenies inferred using the L and S segment differed in regard to the placement of JMSV strains recovered from S. vagrans in Washington ( Figure S3). For host cyt b, S. trowbridgii and S. vagrans were supported as monophyletic; however, S. monticola was shown to be paraphyletic, with a coastally distributed clade, including the sole S. bairdi sample, that is sister to S. palustris, and a larger continental clade that is composed of northern and southern subclades containing viruses from the NC and SC clades, respectively ( Figure 2).

Phylogenetic Analysis
Significant values for recombination analyses differed among segment and test used but identified no regions of intragenic rearrangement in any segment analyzed. The JMSV strains showed a pattern of diversification with defined clades that closely mapped to geographic regions similar to that seen in the phylogeographic structure of the hosts, with the exception of the virus recovered from S. palustris, which was previously identified as Fox Creek virus ( [47]; Figure 1). Geographic regions correspond to Northern Continental (NC) and Southern Continental (SC) clades composed of viruses recovered strictly from S. monticola. The NC clade contained JMSV strains from British Columbia and Yukon Territory, as well as Alaska, while the SC clade was found in New Mexico and Colorado. A third clade was recovered from the Pacific coast (PC) which, in comparison to the NC and SC clades, was hosted by several species of the S. vagrans complex, including S. vagrans, S. trowbridgii, S. bairdi, and S. palustris. Phylogenies inferred using the L and S segment differed in regard to the placement of JMSV strains recovered from S. vagrans in Washington ( Figure  S3). For host cyt b, S. trowbridgii and S. vagrans were supported as monophyletic; however, S. monticola was shown to be paraphyletic, with a coastally distributed clade, including the sole S. bairdi sample, that is sister to S. palustris, and a larger continental clade that is composed of northern and southern subclades containing viruses from the NC and SC clades, respectively ( Figure 2).   Table S2. YT, Yukon Territory. GenBank accession numbers for the L-segment sequences are provided in Table  S2.  Table S2.

Population Demographics
Tajima's D statistic for all three viral genomic segments, as well as host cyt b, were −3.8, −4.5, −3.8, and −3.1 for the S, M, L segments and host cyt b, respectively, with all values being highly significant (P < 0.01). Overall mean diversity, as computed in MEGA using a Kimura 2-parameter model, was 0.256. This relatively high level of nucleotide diversity, common in RNA viruses with high mutation rates, is driven mainly by the PC clade and between group divergence. Within clade  Table S2.

Population Demographics
Tajima's D statistic for all three viral genomic segments, as well as host cyt b, were −3.8, −4.5, −3.8, and −3.1 for the S, M, L segments and host cyt b, respectively, with all values being highly significant (P < 0.01). Overall mean diversity, as computed in MEGA using a Kimura 2-parameter model, was 0.256. This relatively high level of nucleotide diversity, common in RNA viruses with high mutation rates, is driven mainly by the PC clade and between group divergence. Within clade distances were highest in the PC clade at 0.196 followed by the NC clade at 0.125. The SC clade had the lowest diversity at 0.053; however, this relatively low value could be an artifact of smaller sample size.

Cophylogeny Tanglegrams
Cophylogeny of the L segment and host tree reflected a pattern of codiversification within continental S. monticola, but a complex pattern of host switching across multiple species within the PC clade. Host phylogeny based on cyt b placed the NC and SC clades as sister lineages forming a larger continental S. monticola group that matched the pattern of diversification seen in the L segment phylogeny (Figure 3). Included in the JMSV L segment PC clade were viruses harbored by S. vagrans, S. trowbridgii, and S. bairdi that corresponded to taxa spread across the host phylogeny. Of note was the placement of S. bairdi within the coastal S. monticola clade, which represented the only sample of JMSV from that clade. Additionally, JMSV recovered from S. trowbridgii was distributed across the JMSV PC clade, yet S. trowbridgii was only distantly related to the S. vagrans complex. Sorex cinereus, the host of Ash River virus, which was used as the outgroup to JMSV in this study, was more closely related to the S. vagrans complex than S. trowbridgii. Cophylogenies based on the M segment ( Figure S2) and S segment ( Figure S3) of JMSV showed similar patterns to those recovered with the L segment. However, phylogeny reconstruction based on a single gene for the hosts, in this case cyt b, can be misleading and should be further tested with additional, independent loci for this complex of shrew species.
Viruses 2019, 11, x FOR PEER REVIEW 7 of 15 distances were highest in the PC clade at 0.196 followed by the NC clade at 0.125. The SC clade had the lowest diversity at 0.053; however, this relatively low value could be an artifact of smaller sample size.

Cophylogeny Tanglegrams
Cophylogeny of the L segment and host tree reflected a pattern of codiversification within continental S. monticola, but a complex pattern of host switching across multiple species within the PC clade. Host phylogeny based on cyt b placed the NC and SC clades as sister lineages forming a larger continental S. monticola group that matched the pattern of diversification seen in the L segment phylogeny (Figure 3). Included in the JMSV L segment PC clade were viruses harbored by S. vagrans, S. trowbridgii, and S. bairdi that corresponded to taxa spread across the host phylogeny. Of note was the placement of S. bairdi within the coastal S. monticola clade, which represented the only sample of JMSV from that clade. Additionally, JMSV recovered from S. trowbridgii was distributed across the JMSV PC clade, yet S. trowbridgii was only distantly related to the S. vagrans complex. Sorex cinereus, the host of Ash River virus, which was used as the outgroup to JMSV in this study, was more closely related to the S. vagrans complex than S. trowbridgii. Cophylogenies based on the M segment ( Figure S2) and S segment ( Figure S3) of JMSV showed similar patterns to those recovered with the L segment. However, phylogeny reconstruction based on a single gene for the hosts, in this case cyt b, can be misleading and should be further tested with additional, independent loci for this complex of shrew species.  In our study, all segments of JMSV supported monophyletic NC and SC clades that mapped to divergent, reciprocally monophyletic S. monticola clades in the host phylogeny and corresponded to the same geographic regions. While the shallow topology (i.e., branching near the terminal tips) seen within JMSV and the NC and SC clades of S. monticola differed and do not map one-to-one, the geographic clades were reciprocally monophyletic and generally supported a pattern of host-parasite codiversification within the continental group.
In contrast, the PC clade of JMSV showed a complex pattern of host switching among sympatric, yet often deeply divergent species of shrews along the Pacific coast. While the PC clade is monophyletic in all segments of JMSV, this clade is distributed across four separate host species, hence indicative of widespread host switching. Of note among the JMSV PC clade was the placement of a virus recovered from S. palustris (MSB144181) that was well supported within the PC clade. The host specimen was, however, collected in the Yukon Territory, well within the range of the NC clade of both JMSV and S. monticola. This pattern of host switching was mirrored by the JMSV strains hosted by S. trowbridgii and S. vagrans that were spread across the PC clade and did not form a single species-specific clade. Virus-virus cophylogenies based on the L and S segments largely mirrored each other with the exception of the placement of two JMSV strains recovered from S. trowbridgii from Washington, which likely was an artifact of sequence coverage for those two segments ( Figure S6). A single virus recovered from a S. vagrans (MSB83395) specimen from Vancouver Island, British Columbia, was not supported as being a member of any of the three geographic clades within JMSV (Figure 3 and Figure S4). It is important to note, however, that these analyses are based on tree topologies that are not fully resolved, resulting in uncertainty regarding sister relationships within both the host and viral trees, which can impact inferences of codiversification and host switching within this system. More sampling is necessary to resolve tree topologies and fully elucidate the evolutionary history of both the shrew hosts and JMSV.

Codivergence and Phylogenetic Reconciliation
We compared metrics that test for codivergence as they fundamentally differ in their methods of calculating tree similarity. The nPH85 statistic was similar for both S and L segments: Their relatively high values, 0.80 and 0.76, respectively, indicated host switching rather than codiversification as the leading pattern of diversification within JMSV. Codiversification previously has been reported as the more prevalent coevolutionary pattern within Bunyavirales [5]. That pattern, however, is not necessarily reflected in the TripL and Trip metrics. For the L segment, the TripL and Trip metrics were 0.41 and 0.44, respectively, indicating that the L segment and host phylogenies shared roughly 60% of triplets. This result differs from the TripL and Trip metrics for the S segment at 0.23 and 0.19, respectively, relating to roughly 80% of shared triplets between the S segment and host phylogenies.
In contrast to the metrics of codivergence between the L or S segment with the host phylogeny, the metrics comparing the L segment to the S segment in some cases show greater similarity between virus and host than between segments of JMSV. While the nPH85 metric for the S and L comparison is less than that for either comparison with the host, it is still relatively elevated at 0.5 indicating that the two segments share roughly half of their internal structure to each other. Also striking is the TripL and Trip scores for the comparison between segments at 0.46 and 0.49, respectively, also indicating that in addition to the variation on internal tree structure, the S and L segment only share about half of their triplets, less than the S segment shares with the host phylogeny. Differences in evolutionary history between the S and L segments of JMSV are reflected by phylogenetic reconciliation for each segment with the host phylogeny ( Figure 4). These segments suggest that a distinctive set of host switching, codivergence, and local extinction events, are necessary to reconcile the virus segment phylogeny with the host phylogeny. This pattern of independence between segments is consistent with a history of reassortment among the L and S segments of JMSV.

Codiversification Processes
Comparative phylogeography of viruses and their associated mammal reservoir hosts can shed light on the processes driving patterns of coevolutionary diversification [2,48] by revealing the role of historical events that shaped contemporary diversity. Contact between divergent hosts may facilitate transmission of viruses to novel hosts, or reassortment of divergent viral components. Comparative phylogeographic studies provide the spatial and temporal foundation necessary for understanding viral evolution, transmission, and disease emergence; these are essential tools for researchers and public health agencies to proactively approach disease emergence and mitigation. In this study, we addressed the evolutionary history that shaped modern diversity within JMSV and associated mammals hosting this virus.
Diversification within JMSV largely reflects the recent biogeographic history of the shrew host species. The phylogeny inferred from cyt b for the S. vagrans complex supports previously reported species designations and relationships [16]. Demboski and Cook [16] recovered substantial geographic structure within S. monticola, identifying distinct clades distributed in northern and southern continental North America, and a third distributed along the Pacific coast. Representative clades in JMSV match the NC and SC clades, respectively, while the PC clade is comprised of viruses

Codiversification Processes
Comparative phylogeography of viruses and their associated mammal reservoir hosts can shed light on the processes driving patterns of coevolutionary diversification [2,48] by revealing the role of historical events that shaped contemporary diversity. Contact between divergent hosts may facilitate transmission of viruses to novel hosts, or reassortment of divergent viral components. Comparative phylogeographic studies provide the spatial and temporal foundation necessary for understanding viral evolution, transmission, and disease emergence; these are essential tools for researchers and public health agencies to proactively approach disease emergence and mitigation. In this study, we addressed the evolutionary history that shaped modern diversity within JMSV and associated mammals hosting this virus.
Diversification within JMSV largely reflects the recent biogeographic history of the shrew host species. The phylogeny inferred from cyt b for the S. vagrans complex supports previously reported species designations and relationships [16]. Demboski and Cook [16] recovered substantial geographic structure within S. monticola, identifying distinct clades distributed in northern and southern continental North America, and a third distributed along the Pacific coast. Representative clades in JMSV match the NC and SC clades, respectively, while the PC clade is comprised of viruses recovered from several other species in the S. vagrans complex, including S. bairdi which falls within the Pacific coastal S. monticola clade. That pattern ostensibly parallels two suggested evolutionary diversification events within the S. vagrans species complex [16]. To date, the absence of JMSV in S. monticola specimens representative of the PC clade (Oregon and Washington) may reflect either true absence or simply low sampling coverage. The validity of S. bairdi as a species separate from coastal S. monticola is questionable and points to other poorly defined species limits in this shrew complex that complicate our assessment [16]. Expanded shrew sampling and viral screening that aims to refine the geographic extent of host limits and viral diversity in western North America are necessary.
The hypothesized initial divergence event within the S. vagrans complex occurred in the Pacific Northwest coast resulting in current inter-species diversity seen within the complex. Subsequent post-glacial expansion followed the Last Glacial Maximum and produced the currently recognized geographic structure within montane shrews (e.g., NC and SC clades) and their close relatives (e.g., Sorex palustris [17]) ( Figure 5). With JMSV largely mirroring the host pattern, codiversification between JMSV and its shrew hosts appears likely within the NC and SC clades. JMSV emergence tentatively can be dated to the initial diversification of the S. vagrans complex ca. 2 MYA [16], with subsequent, possibly multiple independent, post-glacial expansion events within S. monticola, and likely S. palustris, during the mid to late Quaternary. Whether the virus recovered from S. palustris within the PC clade is the result of a recent host-switching event, is unclear, and more sampling is necessary to refine our understanding of the evolutionary history of JMSV within S. palustris. The single virus recovered from a S. vagrans on Vancouver Island does not align within a defined mainland clade in the S segment and exists on a long branch in the L segment. This finding raises the question of whether there is an endemic insular clade of JMSV similar to that identified for insular S. monticola [49]. Highly negative Tajima's D values for all three viral segments, coupled with observed patterns of sequence divergence centered in the hypothesized source population of the PC clade, strengthens the hypothesis of post-glacial expansion for JMSV ( Figure 5B). While the timing of orthohantavirus diversification remains elusive [7,50,51], this shared pattern of diversification, both spatially and temporally, between JMSV and its mammalian hosts is consistent with a cophylogeographic history that is much deeper than several thousand years.
While phylogenies for JMSV and its host species largely mirror each other in deeper phylogenetic structure, codiversification is only partially responsible for contemporary diversity. Elevated codivergence metrics calculated between trees suggest host switching also plays an important role. Geoghegan and colleagues [5] applied the nPH85 metric to family level phylogenies for several RNA and DNA viruses. Our results for this metric largely match the general pattern seen for other RNA viruses [5]. When comparing topologies between the host phylogeny and either the S or L segment, we calculated an nPH85 value of 0.8 and 0.76, indicating scant codivergence, which is consistent with the values seen at the family level. Furthermore, when calculated for the comparison between the L and S segments, we obtained a value of 0.5, indicating a mix of codivergence and host switching between each segment. This is in contrast with the tanglegram for the L and S segment comparison ( Figure S5) which indicated the phylogenies largely mirrored each other. However, the level of similarity between tree topology and groupings of terminal taxa do not necessarily tell the same story as indicated by drastically different TripL and Trip values compared to nPH85. Internal branching structure is driving the difference in codivergence metrics calculated between phylogenies inferred from the L and S segments. However, our sampling of the L segment is more complete in terms of number of specimens and coverage of the segment. This fact, coupled with our phylogenetic reconciliation analysis, suggests that in addition to a complex history of post-glacial codivergence and local host switching between JMSV and the S. vagrans complex as a whole, there is additional complexity due to independent evolutionary histories associated with distinct viral segments. Increased sampling and full-length genomic sequencing would help test whether incongruity between the S and L segments, both tree topology and similarity metrics, reflects deeper phylogenetic patterns or is merely an artifact of sampling and sequencing bias.
necessary to refine our understanding of the evolutionary history of JMSV within S. palustris. The single virus recovered from a S. vagrans on Vancouver Island does not align within a defined mainland clade in the S segment and exists on a long branch in the L segment. This finding raises the question of whether there is an endemic insular clade of JMSV similar to that identified for insular S. monticola [49]. Highly negative Tajima's D values for all three viral segments, coupled with observed patterns of sequence divergence centered in the hypothesized source population of the PC clade, strengthens the hypothesis of post-glacial expansion for JMSV ( Figure 5B). While the timing of

Viral Reassortment
Viral reassortment is possible for viruses with segmented genomes and can be a catalyst for driving diversification and pathogenesis. The 2009 influenza outbreak is a prime example of reassortment among multiple divergent strains resulting in a pandemic [13]. Reassortment produces unique combinations of viral segments that have the potential to influence pathogenicity due to presentation of novel virions to an immunologically naïve population. Co-circulation of divergent viruses within a single cell is hypothetically necessary for reassortment and calls for a more detailed understanding of the range of ecological circumstances that can lead to viral switching between hosts if we hope to predict disease emergence [2].
Historically, reassortment events within orthohantaviruses were thought to be relatively rare compared to other members of Bunyavirales due to method of transmission (i.e., direct host-host contact versus arthropod transmission, respectively) [12]. However, an increasing number of studies of orthohantaviruses have shown instances of reassortment [53][54][55][56], both ancient and modern. The extent of reassortment and its contribution to modern-day orthohantavirus diversity is a new avenue of research and there is still much to learn. Implicit in orthohantavirus reassortment events is the necessity for contact between hosts for transmission of divergent viruses, or what are called spillover events. For instance, there is evidence in Finland for two distinct strains of Puumala virus co-circulating with active and ongoing reassortment resulting from contact, both modern and historic, between divergent populations of the host species, M. glareolus [54]. A more contemporary example of co-circulation was reported in Belgium with two deeply divergent strains of Hantaviridae co-circulating in the European mole, Talpa europaea [57]. That system is unique in that multiple strains of virus have been recovered from a single individual host sample, potentially representing a prime situation for reassortment. Such scenarios, where reassortment and host-switching events can occur, highlight that studies that expand our knowledge of both geographic and host range are crucial to gaining a better understanding of the role of host ecology and evolutionary history in viral spread, emergence, and pathogenesis.
Incongruence in tree topologies between the L and S segments likely reflects host-switching and reassortment in the evolutionary history of JMSV. The large geographic distance between the NC and SC clades (Canada/Alaska and New Mexico/Colorado, respectively) containing the possible reassortant strain suggests that this event likely predates the post-glacial expansion of S. monticola. However, the lack of sampling spanning the distance between the NC and SC clades, minimal sampling for the M segment to date, and only a single instance of reassortment, precludes full elucidation of its role in the evolutionary history of JMSV. To attain a more complete history of JMSV, and shrew-borne orthohantaviruses in general, a comprehensive sequence dataset of all three genomic segments is needed, spanning the breadth of host diversity. Klempa showed that reassortment among orthohantaviruses is more prevalent than originally believed [10], however, the study of orthohantavirus reassortment is relatively under-explored and the extent of this process in orthohantavirus diversification remains largely unknown.
Shi and colleagues [58] provided a better understanding of the deeper phylogenetic and evolutionary history responsible for the current diversity in vertebrate viruses at the order and family level, revealing that an overall trend of codivergence is coupled with a complex history of host switching between distantly related taxa. Bennett and co-workers [47] showed that these trends of host switching and codivergence hold true within a single virus genus, Orthohantavirus, in their exploration of the relationships and phylogeographic history of several strains of orthohantavirus. In addition, Torres-Pérez and colleagues [7] examined the phylogeographic history and patterns of divergence within a single strain and host, Andes virus hosted by the South American rodent, Oligoryzomys longicaudatus. Their results revealed that while there was overall similarity in spatial structure between virus and host phylogenies, the timing of diversification was incongruent. That research highlights the difficulty of accurately dating the evolution of hantaviruses, with estimates ranging broadly from several thousand to several million years before present [50,[59][60][61]. Our study of a single orthohantavirus that is shared among multiple soricid host species begins to explore the role of host history across a temporal scale that spans across population-level processes to much deeper evolutionary events.
JMSV does not show a history of strict codiversification, but rather multiple host-switching events at both broad and fine geographic scales, and at least one instance of reassortment of divergent strains. More sampling is necessary to elucidate the evolutionary history of JMSV within S. palustris, S. bairdi, and other close relatives of this shrew complex, and would help resolve uncertainty of tree topology which limits our ability to fully understand this system. Nonetheless, it is evident that JMSV has a complex and relatively deep evolutionary history in North America. That history remains complicated by uncertainty of host taxonomy, such as polyphyletic assemblages in both S. monticola and S. palustris [16,17]. Such uncertainty illuminates the necessity for a solid understanding of host relationships and history when discerning the evolutionary history of viruses and parasites in general.

Supplementary Materials:
The following are available online at http://www.mdpi.com/1999-4915/11/7/637/s1. Figure S1: Geographic sampling for JMSV L segment and host cytochrome b, Figure S2: Maximum likelihood phylogeny for JMSV M segment, Figure S3: Maximum likelihood phylogeny for JMSV S segment, Figure S4: Tanglegram for host cytochrome b and JMSV M segment, Figure S5: Tanglegram for Sorex vagrans complex on the left, and Jemez Springs S segment on the right, Figure S6: Tanglegram for Jemez Springs L segment on the left and the S segment on the right, Table S1: Oligonucleotide primers for amplification of JMSV, Table S2: GenBank accession numbers for sequences used in this study, File S1: Sequence alignments and newick format phylogenies used in this study.