Fish conservation in the land of steppe and sky: Evolutionarily significant units of threatened salmonid species in Mongolia mirror major river basins

Abstract Mongolia's salmonids are suffering extensive population declines; thus, more comprehensive fisheries management and conservation strategies are required. To assist with their development, a better understanding of the genetic structure and diversity of these threatened species would allow a more targeted approach for preserving genetic variation and ultimately improve long‐term species recoveries. It is hypothesized that the unfragmented river basins that have persisted across Mongolia provide unobstructed connectivity for resident salmonid species. Thus, genetic structure is expected to be primarily segregated between major river basins. We tested this hypothesis by investigating the population structure for three salmonid genera (Hucho, Brachymystax and Thymallus) using different genetic markers to identify evolutionarily significant units (ESUs) and priority rivers to focus conservation efforts. Fish were assigned to separate ESUs when the combined evidence of mitochondrial and nuclear data indicated genetic isolation. Hucho taimen exhibited a dichotomous population structure forming two ESUs, with five priority rivers. Within the Brachymystax genus, there were three B. lenokESUs and one B. tumensisESU, along with six priority rivers. While B. tumensiswas confirmed to display divergent mtDNA haplotypes, haplotype sharing between these two congeneric species was also identified. For T. baicalensis,only a single ESU was assigned, with five priority rivers identified plus Lake Hovsgol. Additionally, we confirmed that T. nigrescens from Lake Hovsgol is a synonym of T. baicalensis. Across all species, the most prominent pattern was strong differentiation among major river basins with low differentiation and weak patterns of isolation by distance within river basins, which corroborated our hypothesis of high within‐basin connectivity across Mongolia. This new genetic information provides authorities the opportunity to distribute resources for management between ESUs while assigning additional protection for the more genetically valuable salmonid rivers so that the greatest adaptive potential within each species can be preserved.


| INTRODUC TI ON
Mongolia's vast river basins include some of the least impacted freshwater ecosystems on the planet (Hofmann et al., 2015). However, across many regions, a multitude of anthropogenic pressures is currently threatening this pristine status, with increasing contamination and degradation of aquatic environments and their resident species (Stubblefield et al., 2005;Karthe et al., 2017). The damage to Mongolia's rivers, streams, and lakes has been a direct result of recent, rapid development, where high rates of urbanization and industrialization have led to growing discharge of poorly treated wastewater, increased industrial pollution and mining contamination, along with rising rates of overgrazing and deforestation in many river basins (Hartwig et al., 2016;Hofmann, Venohr, Behrendt, & Opitz, 2010;. However, the main driver of many native fish population declines, including a number of salmonid species that have suffered regional losses suspected of being up to 50% in recent decades, has been intensifying fishing activities, which has continued to gain rapid popularity across the country (Chandra, Gilroy, Purevdorj, & Erdenebat, 2005;Hogan & Jensen, 2013;Kaus, 2018;Ocock et al., 2006). In order to help mitigate these declines, improvements to the current fisheries management strategies are required, with an essential first step being the identification of ecologically meaningful management units across the wider distribution of each target species so authorities and policy makers can better understand the functional scale of the threatened populations that they are trying to manage and conserve (Funk, McKay, Hohenlohe, & Allendorf, 2012). Evolutionarily significant units (ESUs) are a common management tool used in conservation biology that involves the identification of intraspecific groups which represent a more biologically meaningful assemblage within a species' geographic distribution. While a number of important factors are typically considered in order to define species' ESUs, including ecologically relevant phenotypic attributes or certain life history traits; one of the more unequivocal techniques that can determine a measurable divergence between isolated groups of conspecifics has been modern genetic methods (Avise, 2000;Bernard et al., 2009;Fraser & Bernatchez, 2001). ESUs can consist of multiple allopatric populations and can cover extensive geographic regions depending on the species and its ecology (Moritz, 1994;Palsbøll, Bérube, & Allendorf, 2006). Units are usually defined based on neutral and sometimes adaptive genetic variation, which represent the effects of both historical spatial processes and environmental selection (Casacci, Barbero, & Balletto, 2014;Crandall, Bininda-Emonds, Mace, & Wayne, 2000;Funk et al., 2012;Moritz, 1994). For the initial demarcation of an ESU, researchers have focused on genetic markers including maternally transmitted, slowly evolving mtDNA, but also biparentally transmitted, quickly evolving microsatellites, as both yield relevant information on complementary spatiotemporal scales (O'Connell & Wright, 1997;Vogler & DeSalle, 1994). The identification of ESUs and genetically distinct populations of threatened and exploited fish stocks is increasingly used in fishery management to ensure that conservation actions and resources can be better matched with biological relevance (Xia, Chen, & Sheng, 2006;Geist, Kolahsa, Gum, & Kuehn, 2009;Escobar, Andrade-López, Farias, & Hrbek, 2015;Zhivotovsky et al., 2015). While ESUs represent the upper hierarchical levels of intraspecific biodiversity, demographically independent groups that harbor an above average genetic variation or are more genetically distinct compared to the rest of the ESU are also important to identify. With recent studies reporting a strong association between alleles at one or a very few genes and a key life history trait in Pacific salmonid species (Hess, Zendt, Matala, & Narum, 2016;Prince et al., 2017), knowledge of these more genetically valuable or priority populations and their geographic extent, i.e. their river system, may be increasingly important to identify for the conservation of other threatened salmonids as well. Such information can assist in designing adequate protection and recovery programs for threatened species, as conservation efforts can thus focus on preserving the ability of natural ecological and evolutionary processes which produce genetic variation capable of sustaining a species long term under future shifting environmental conditions (Petit & Mousadik, 1998;Waples & Lindley, 2018).
Mongolia's salmonids species from the genera Hucho, Brachymystax, and Thymallus (Family Salmonidae) live in sympatry throughout the country's two major river basins. While there are still some remote rivers systems holding robust numbers of these species, widespread declines have continued throughout much of their range both in Mongolia and neighboring Russia and China (Hogan & Jensen, 2013;Ocock et al., 2006). The three main targets within the recreational fishery include the Siberian taimen (Hucho taimen, Pallas 1773), the sharp-snouted lenok (Brachymystax lenok, Pallas 1773), and the Baikal grayling (Thymallus baicalensis, Dybowski 1874). Two additional species, the blunt-snouted lenok (B. tumensis, Mori 1930) and the Hovsgol grayling (T. nigrescens, Dorogostaisky 1923), are also commonly caught and killed and are thus also suffering a similar fate across their more restricted distributions in Mongolia including the Onon River (Amur Basin) and Lake Hovsgol (Selenge Basin), respectively.
Phylogeographic research on these salmonid species has revealed population genetic structure across various geographic scales Kuang, Tong, Xu, Sun, and Yin, (2010) (Froufe, Alekseyev, Knizhin, & Weiss, 2005;Maric et al., 2014). A past population bottleneck has been proposed as the likely cause of such low genetic diversity within the world's largest salmonid species, which has occurred prior to a relatively recent range expansion during a period of hydrological exchange between neighboring Siberian river basins (Froufe, Alekseyev, Knizhin, Alexandrino, & Weiss, 2003;Grosswald, 1998;Holčík, Hensel, Nieslanik, & Skacel, 1988). It has been hypothesized that if the population bottleneck occurred after the predicted range expansion, then multiple genetic lineages would have likely occurred, which appears to not be the case (Froufe et al., 2003). However, further detailed genetic research focused on H. taimen populations in the understudied Mongolian river systems may yet identify putative genetically independent lineages.
The Brachymystax genus is made up of three recognized species with two of these residing in Mongolia including B. lenok (sharpsnouted lenok), the most widely distributed and commonly captured species (Kaus, 2018) and B. tumensis (blunt-snouted lenok), a similar looking species which is found in more fragmented populations in the Onon River (Amur River Basin) (Bogutskaya & Naseka, 1996;Ma et al., 2007;Froufe, Alekseyev, Alexandrino, & Weiss, 2008;Xing et al., 2015). To date, both broad-scale phylogroups and intrabasin genetic structuring have been identified for B. lenok between the major river basins across northern Asia/Siberia and within Chinese rivers, respectively (Froufe et al., 2008(Froufe et al., , 2003Maric et al., 2014;Xia et al., 2006). However, the Mongolian populations of this species remain largely unstudied in detail and it is yet to be determined whether B. lenok displays intrabasin genetic structure outside of the highly fragmented Chinese basins. In addition, while the species status of the sympatric B. tumensis in Mongolia has created confusion among ichthyologists in the past, in other regions, the blunt-snouted lenok has been demonstrated to have had a clear genetic divergence and thus been reported to have undergone a long and independent evolutionary history (Froufe et al., 2008(Froufe et al., , 2003Shed'ko, 1996). However, due to the minimal amount of ichthyological research in Mongolia, the question has remained whether the resident blunt-snouted lenok populations are in fact B. tumensis, an intraspecific form of B. lenok (as has been previously accepted), or a different species altogether. To date, only one lenok species (B. lenok) has been recorded on the country's official fish species list, including the country's Red List of Fishes (2006). Thus, this issue should be explicitly addressed using modern genetic methods in order to confirm the status of the Mongolian Brachymystax species and populations.
In Mongolia, there are five Thymallus species currently listed, with T. baicalensis being recently confirmed as the species inhabiting the Selenge River Basin  after it was previously thought to be T. arcticus (Pallas, 1776). However, it remains unclear whether any further Thymallus species reside in the expansive Selenge River system as has been suggested (Kottelat, 2006), or whether there is clear genetic sub-structuring displayed by this species that would need to be considered in more comprehensive fisheries management plans. While the status of the Mongolian grayling (T. brevirostris, Kessler, 1879), the Amur grayling (T. grubii, Dybowski, 1869), and the upper Yenisei grayling (T. svetovidovi, Knizhin & Weiss, 2009) is clear, further clarification is required to determine whether the Hovsgol grayling (T. nigrescens, Dorogostaisky, 1923) from Lake Hovsgol in the central north of Mongolia represents a unique and independent species or not, as it has been treated as a separate species by some authors (Berg, 1962;Bogutskaya & Naseka, 2004;Pivnička & Hensel, 1978), but not by others (Froufe, Alekseyev et al., 2005;Knizhin, Weiss, & Sušnik, 2006;Koskinen et al., 2002).
In order to clarify the status of the aforementioned species, delineate their ESUs and identify genetically diverse or differentiated intraspecific populations that should be conservation priorities, a combination of mtDNA sequencing and microsatellite marker analyses was conducted on sampled individuals from each species across their entire Mongolian distributions including the upper Yenisei, the Selenge, and upper Amur river basins. It is hypothesized that the unfragmented river basins in Mongolia, which are unique for such large, boreal systems in the world, have allowed for unobstructed connectivity and thus unrestricted movement and intergenerational gene flow over large spatial scales, which has ultimately resulted in the genetic structure of these species being primarily segregated between river basins with minimal differentiation existing within each river basin. Our aim was to also shed light on the phylogenetic relationships and species status of B. tumensis and T. nigrescens using genetic techniques. This research can provide a more detailed understanding of the genetic structure of these threatened salmonid species in Mongolia, while providing authorities with an improved ecological understanding in order to develop more comprehensive management strategies and better protect rivers holding genetically valuable populations that can help to safeguard the evolutionary potential and adaptability of these species for future climatic challenges.

| Study area
Mongolia contains the most upstream regions of two major Eurasian drainages ( Figure 1). The Selenge River Basin, with a catchment that covers most of northcentral Mongolia, flows north into Siberia and forms the main inflow to Lake Baikal. From Lake Baikal, water continues via the Angara River into the Yenisei River which ultimately discharges into the Arctic Ocean. The Shishged River is a major trib-  (Froufe et al., 2003;Grosswald, 1998Grosswald, , 1999Koskinen et al., 2002)

| Field sampling
The current sampling design was intended to capture the complete genetic diversity of the investigated species from across their entire Mongolian distributions. Fish were sampled from 19 rivers within the Yenisei, Selenge, and Amur basins in 2011 and 2012 (Table 1, Figure 1). Rivers were selected due to the reported species present, the isolation by river distance, and river accessibility. Sample sites were typically in the upper reaches, but due to the low abundances of certain species in a number of rivers, it was necessary to cover tens of kilometers in order to collect sufficient numbers of samples.
We collected fin clips from a total of 127 H. taimen from seven rivers, 371 B. lenok from 18 rivers, and 274 T. baicalensis from 11 rivers. We also collected samples of B tumensis from the Amur basin (12 individuals from 3 rivers) and T. nigrescens from Lake Hovsgol (15 individuals). Fish were caught using backpack electrofishing units (Hans Grassel GmbH, Germany; Type ELT 60) and angling by researchers and international fishing guides, with all of these individuals being released alive. Several samples were also collected from fish caught by local recreational anglers. Fin clips were placed into 96% ethanol for transportation and storage until analysis at the Helmholtz Centre for Environmental Research (UFZ) in Halle (Saale), Germany.

| Genotyping
DNA was extracted from fin clips using the DNeasy Blood and Tissue kits (QIAGEN, Hilden, Germany) following the manufacturer's instructions. We sequenced the control region ("D-loop") of mitochondrial DNA from a total of 31 H. taimen that were sampled from across seven rivers, a total of 35 B. lenok from across 12 rivers, and a total of 11 B. tumensis from two rivers, as well as a total of 20 T. baicalensis from across six rivers and three T. nigrescens from Lake Hovsgol using primers LRBT-25 and LRBT-1195 (Uiblein, Jagsch, Honsig-Erlenburg, & Weiss, 2001). Details of the sequencing reaction are given in the Supplementary Material.
All H. taimen, Brachymystax, and Thymallus samples were genotyped at eleven, eight, and eight microsatellite loci, respectively (Supporting information Table S1). A small number of loci produced multiple bands which could be consistently scored as independent loci, one in Brachymystax (BleTri4) and two in H. taimen (BleTri4 and BleTet6). We used a PCR protocol with CAG/M13R-tagged forward primers and GTTT-"pigtailed" reverse primers following Schuelke (2000). Primer sequences and details of the PCR protocol are given in the Supplementary Material.

| Data analysis
Following preliminary analysis, we combined individuals from adjacent collection sites within a specific river when sample sizes were particularly low; thus, we refer to species "populations" even F I G U R E 1 Displays the sample locations across northern Mongolia. The major river basins shown include the Yenisei/Selenge River Basin (Arctic Ocean drainage in red shading), the Amur River Basin (Pacific drainage in blue shading), and the Central Asian basin (yellow shading). Only a very upper tributary of the Yenisei River (Shishged River), which is downstream of Lake Baikal, is located in Mongolian territory and is labeled as Y1. Hucho taimen samples were collected from sites Y1, S1, S3, S6, S10, A2, A3, and A5; Brachymystax lenok samples were collected from all sample points; B. tumensis were sampled from A2 -A4; Thymallus baicalensis were samples from S1-S12 (excluding S5), and T. nigrescens was sampled from S5 only TA B L E 1 Genetic diversity of three salmonid genera sampled from the Yenisei, Selenge, and Amur rivers basins in Mongolia in 2011 and 2012 if sample groups represented considerable parts of the same river.
Full analyses were then carried out separately for each of the three salmonid genera collected across Mongolia. Mitochondrial DNA data and additional sequences acquired from GenBank (for Brachymystax spp. and H. taimen) were aligned using Geneious® Pro 5.6.7 (Kearse et al., 2012) and the build-in multiple alignment option. Haplotype networks were obtained by using a median-joining algorithm (Bandelt, Forster, & Rohl, 1999)   From the microsatellite data sets, we calculated descriptors of population genetic variation, that is, the number of alleles (A), allelic richness (A R ), expected heterozygosity (H e ) as well as the inbreeding coefficient (F IS ) and its significance (p-value) using FSTAT v2.9.3.2 (Goudet, 2001 (Francis, 2016). When multiple clusters were found, we reanalyzed these clusters separately as the STRUCTURE software was sensitive to hierarchical population structure. Population differentiation was quantified with hierarchical analyses of molecular variance (AMOVA) conducted in GeneAlEx 6.5 (Peakall & Smouse, 2006. We tested for isolation by distance within basins, that is, a correlation between genetic differentiation and distance along the river with Mantel tests in R (R Core Team, 2013). River distance was estimated between hydrologically connected sites by tracing the main river channel and using the measuring ruler "path" in TA B L E 1 (Continued) based on allelic richness thus correcting for unequal sample sizes.
The contribution of populations to total species diversity was partitioned into two components: the diversity of individuals within that river and their differentiation from other rivers. Although genetic diversity/differentiation are common metrics used to identify priority populations within a species distribution, it is not the only definition that has been used in conservation biology studies of threatened species, with population viability, highest risk and/or greatest ecological consequences following extinction having also been used as criteria to prioritize conservation resources (Allendorf et al., 2002).
In any case, within the scope of the current research, only genetic metrics have been used to delineate priority populations within the different salmonid species' ESUs. However, while this detailed genetic information has provided initial conservation priorities, ultimately these data can later be combined with ecological and demographic population assessments to further enhance salmonid species' management strategies in Mongolia.

| H. taimen mitochondrial and nuclear markers
Four mitochondrial haplotypes were identified within the sampled H. taimen (Table 1, Figure 2). These could be assembled into two main groups, separated by four mutations.  Figures 3, S1a). No further genetic structure was evident when each basin cluster was analyzed again separately ( Figure S1b and c). These clusters were also supported by the principal component analysis (PCoA), where axis one explained 31.7% of the variation and axis two explained 7.7% of the variation ( Figure S6). The AMOVA for H. taimen indicated that 29% of the genetic variance was partitioned among basins, 1% among rivers within basins, and the rest residing within rivers (Table S5a).
Separate analyses for the Selenge (incl. Shishged) and Amur basins revealed F ST = 0.027 (p = 0.001) and F ST = 0.052 (p = 0.008), respectively (  (2005), with haplotypes detected in this study labeled using existing haplotype names (Froufe, Alekseyev et al., 2005, Figure 2c). Haplotypes found in this study are underlines, while new haplotypes for the species are denoted with an asterisk (*). Note that some of the previously identified haplotypes (Froufe et al., 2003;Froufe, Alekseyev et al., 2005) collapsed into a single haplotype because the total alignment was shorter F I G U R E 3 Bayesian cluster analysis with STRUCTURE for the microsatellite data of H. taimen sampled in eight rivers from the Yenisei (Y1), Selenge (S1, S3, S6, S10), and Amur basins (A2, A3, A5) across Mongolia. Individual proportional membership is shown for K = 2 as two genetic clusters were identified by the Evanno plots ( Figure S1). Each identified cluster was again run separately and both displayed K = 1 unequal sizes to be compared without bias and the total genetic diversity to be partitioned into the contribution of genetic diversity within the population and the contribution of genetic differentiation of a population within an ESU (Figure 4, Table 1). For H. taimen from three rivers in the Selenge (Shishged, Delgermoron, and Eg-Uur) and two in the Amur (Onon and Khalkhin), we identified increased genetic diversity and thus were recognized as priority rivers (Figure 4a and b).

| Brachymystax spp. mitochondrial and nuclear markers
A total of eight mtDNA haplotypes were identified within the sampled Brachymystax individuals (Table 1, (Table S3) and the PCoA ( Figure S7). The AMOVA results for B. lenok indicated that 16% of the genetic variance was among basins, 2% among rivers within basins, and the rest residing within rivers (Table S5e) Figure S9a) and the Amur basin (r = 0.76, p = 0.045; Figure S9b). For B. lenok, five priority rivers were identified in the Selenge (Shishged, Delgermoron, Humen, Orkhon, and Tuul; Figure 4c) and two in the Amur (Onon and Kherlen; Figure 4d).

| Thymallus spp. mitochondrial and nuclear markers
Nine mtDNA haplotypes were found among five Thymallus species. The haplotype network showed four distinct groups (Figure 7).  (Table S4). T. baicalensis across the Selenge basin showed no significant pattern of isolation by distance (r = 0.22, p = 0.12, Figure S10). For T. baicalensis (including T. nigrescens), six rivers were considered to be priorities due to the above average F I G U R E 6 STRUCTURE analyses of the microsatellite data for the Brachymystax genus (including both B. lenok and B. tumensis) collected from 19 rivers across the Yenisei (Y1), Selenge (S1-S13), and Amur river basins (A1-A5) in Mongolia. When all samples were included in the analysis, two genetic clusters were identified according to the Evanno plots (top; Figure S2a). These two clusters were further analyzed separately (second row), with the results from the "orange cluster" (Selenge basin) yielding no further genetic structure ( Figure S2b), while within the "green cluster," additional sub-structuring was identified ( Figure S2c). Upon further analysis, no substructure of the "purple cluster" was identified (results not shown). However, within the "green cluster," two genetically distinct populations were clearly displayed ( Figure S2d) Zelter, Huder, and Lake Hovsgol; Figure 4e). In most cases, total diversity was determined by high diversity contributions rather than differentiation contributions, in line with low within-basin divergence.  (Hoban et al., 2013). Such an oversight can have substantial consequences for a species, as the loss of genetic variation can reduce the evolutionary potential at both a population and species level (Barrett & Schluter, 2008;Keller & Waller, 2002;Rivers, Brummitt, Nic Lughadh, & Meagher, 2014). For three taxa of salmonids with high conservation concern in Mongolia, we found that population structure was primarily segregated between major river basins with largely matching patterns between mitochondrial and nuclear genomes. Although exact patterns were not completely concordant among species, we identified strong genetic differentiation among basins but rather weak differentiation within basins. B. lenok was the only species to show a clear pattern of within-basin isolation by distance. Furthermore, patterns of diversity and differentiation allowed for the identification of conservation priority rivers across Mongolia's major basins, with the results indicating that some rivers are valuable for two or more of the sampled salmonid species making them genetic hotspots. Therefore, new management strategies need to recognize the importance of understanding and incorporating genetic diversity and differentiation patterns so a more targeted approach can be developed in an attempt to retain the maximum genetic variation across intraspecific groups and maintain the highest adaptive potential of the focus species.

| Genetic population structure and priorities for conservation of H. taimen
This study has demonstrated that Mongolian H. taimen represents the most upstream extent of the two previously identified major phylogroups, which means that two independent ESUs need to be recognized in conservation and management efforts in the country ( Figure 8a). The Selenge and Shishged ESU forms part of the western phylogroup that consists of the greater Yenisei, Khatanga, Volga, and Ob river basin H. taimen, while the Onon and Kherlen ESU is part of the eastern Amur phylogroup together with Lena basin H. taimen (Froufe, Alekseyev et al., 2005;Maric et al., 2014).
Certain ecological traits such as the H. taimen's large body size and propensity of mature individuals to move and disperse extensive distances particularly during the spawning season (Holčík et al., 1988;Matveyev, Pronin, Samusenok, & Bronte, 1998;Jensen et al., 2009;Gilroy et al., 2010; has likely contributed to the minimal genetic structure found in this long-lived species. Similar patterns of negligible genetic structure across large geographic scales have also been reported in other large-bodied freshwater fishes, which are also known to move extensive distances during their lifetimes (Ferreira et al., 2017;So, Houdt, & Volckaert, 2006;Stepien, Murphy, Lohner, Sepulveda-Villet, & Haponski, 2009). Resident H. taimen in the Shishged River coupled with individuals from the Delgermoron, Eg-Uur, Onon, and Khalkhin rivers collectively represent the most genetically diverse populations within the two ESUs identified in Mongolia ( Figure 8a). Incidentally, these rivers also are known to hold some of the last remaining robust H. taimen populations in Mongolia and thus their protection will be critical, for conserving not only their genetic diversity, but for the persistence of the entire species.

| Genetic population structure and conservation priorities for Brachymystax spp.
For the two Brachymystax species residing in Mongolian rivers, the genetic analysis revealed a total of four distinct ESUs (Figure 8b). We identified three allopatric ESUs belonging to B. lenok from the Selenge, the Shishged and Amur basins, and confirmed B. tumensis from the Amur as a separate, sympatric species and distinct ESU. The Selenge ESU formed an exclusive genetic group based on mtDNA and nuclear markers, which would likely extend as far as Lake Baikal downstream considering the findings of Froufe et al. (2008). Such genetic divergence can be attributed to the reported prior isolation of the Selenge basin and Lake Baikal from the Yenisei and Amur basins approximately half a million years ago (Mats et al., 2001(Mats et al., , 2001. This unique B. lenok phylogroup is genetically distinct and thus highly valuable in the context of genetic conservation for the species. Incidentally, this group has also been the focus of increased scientific research on B. lenok including studies on their feeding ecology (Olson, Jensen, & Hrabik, 2016), thermal tolerances (Hartman & Jensen, 2016), lotic and lentic growth comparisons (Tsogtsaikhan et al., 2017), and seasonal movements (Kaus, Büttner, Karthe, Schäffer, & Borchardt, 2017).
The other two B. lenok ESUs in the Shishged River and Amur basin displayed no distinction at the mtDNA level but were genetically separated from each other according to nuclear microsatellite markers. Thus together with their geographic isolation, their status as separate ESUs was justified. The shared mtDNA haplotype that was found between these ESUs highlighted the relatively recent divergence of these populations, while supporting the hypothesis of the late Pleistocene hydrological connectivity between the Amur and Yenisei basins via the Lena River (Froufe et al., 2008(Froufe et al., , 2003Grosswald, 1998). However, despite this shared haplotype, most of the genetic differentiation for B. lenok was distributed among basins (Froufe et al., 2008;Liu, Kunag, Tong, & Yin, 2011;Xia et al., 2006), which indicated large-scale, intrabasin gene flow within these vast, unfragmented river systems. However, B. lenok was the only species to demonstrate isolation by distance within both river basins, which is in line with the expectation of a reduced dispersal ability compared to the larger sized H. taimen (Yoon et al., 2014;Gilroy et al., 2010;Kaus et al., 2017).
The current results additionally demonstrate that the sympatric populations of B. lenok and B. tumensis from the Amur basin are genetically highly divergent. Thus, B. tumensis represents a fourth ESU for the Brachymystax genus in Mongolia. While natural hybrids have been identified between these two species in regions of sympatry, there has been no evidence of shared haplotypes ever occurring (Froufe et al., 2008;Ma & Jiang, 2007).
However, in contrast to these previous studies, strong evidence was found for nuclear introgression from B. lenok into B. tumensis.
Although this indicates incomplete reproductive isolation, there was only one first generation hybrid identified in the low number of samples collected, thus suggesting there is, at least, a certain level of mitochondrial introgression still occurring between these species in this region of sympatry. This rarity of mixed ancestry in general indicates that hybridization is infrequent or was largely an ancient event. Haplotype sharing could principally be also due to shared ancestral polymorphism, but hybridization appears more likely to be the case, as this is the first such observation reported in these species. Hybridization between congeneric fish is common especially after secondary contact and molecular markers are highly suited to test specific hypotheses (Hänfling, Bolton, Harley, & Carvalho, 2005 and Lake Markakol (Kazakhstan), respectively (Alekseev & Osinov, 2006;Ma et al., 2007). Consequently, these names are regarded as invalid and thus a new scientific name for B. tumensis should be assigned (Froufe et al., 2008;Ma et al., 2007).

| Genetic population structure and conservation priorities for Thymallus spp.
In contrast to common ecological characteristics of the Thymallus genus such as strong natal homing tendencies and poor dispersal abilities, T. baicalensis in the Selenge basin has shown no evidence of genetic structuring among the sampled rivers, which is likely due to not only the comparatively smaller geographic scales but also the high hydrological connectivity that has persisted across the basin.
Thus, T. baicalensis in Mongolia represents a single ESU including individuals from Lake Hovsgol (Figure 8c). As a result, this finding has flow on implications for the species status of T. nigrescens from Lake Hovsgol. While some authors have recognized it as an independent species based on both morphological and biological indices including number of gill rakers and pyloric caeca (Berg, 1962;Bogutskaya & Naseka, 2004;Pivnička & Hensel, 1978), other authorities have expressed the need for additional analyses or have already outright disregarded T. nigrescens as its own species (Knizhin et al., 2006;Koskinen et al., 2002;Weiss et al., 2007). The result from the current research supports the opinion that T. nigrescens is not genetically distinct from T. baicalensis and that these two putative species are in fact synonyms. The lack of genetic distinction detected by the mtDNA marker analysis suggests there is either a significant amount of contemporary gene flow between Lake Hovsgol and Selenge basin inhabitants via the Eg-Uur River or else it has ceased only recently.
The morphological differences displayed by individuals that inhabit Lake Hovsgol, including significant differences in the length-weight and age-length relationships compared to T. baicalensis sampled from riverine environments (Tsogtsaikhan et al., 2017), are likely due to the high ecological flexibility and phenotypic plasticity of this genus, which has previously caused confusion between intraspecific forms in Lake Baikal (i.e., black and white Baikal graylings, Knizhin et al., 2006). However, the contrib analysis and the pairwise F ST values still indicated that individuals sampled from Lake Hovsgol, while not a separate species, displayed genetic differentiation from each other, which justifies the Lake Hovsgol population as a priority within the T. baicalensis ESU. Furthermore, our analysis assessed largely neutral genetic variation and does not exclude the possibility that the selection has affected ecologically important genetic variation. Therefore, the Lake Hovsgol population should still be recognized as a unique intraspecific group that should be afforded adequate conservation and protection efforts to mitigate the growing number of impacts in the region including overfishing, increased pollution, and climate change, which have been reported to be increasingly impacting this ancient lake (Ahrenstorff, Jensen, Weidel, Mendsaikhan, & Hrabik, 2012;Free, Jensen, & Mendsaikhan, 2016).
For T. baicalensis, only minimal differences in the genetic diversity and differentiation contributions were detected across the Selenge ESU. However, in addition to the Lake Hovsgol population, T. baicalensis from five other sampled rivers displayed above average genetic diversity with the Zelter and Huder, and Orkhon and Tuul, appearing to share the same proportion of genetic contribution (Figures 4e and   8c). This is likely due to the close proximity of the river confluences, and thus, a substantial amount of genetic exchange is expected to have caused this similarity. The Delgermoron was the fifth population that was identified as having an elevated genetic diversity component compared to populations from other rivers, thus also making it a priority river for the conservation of this species in Mongolia.

| Patterns across species and implications for broader conservation strategies
While a broad understanding of population genetics is crucial for threatened species management, neutral marker patterns represent only one fundamental aspect for defining conservation objectives, and thus, a range of other biological, ecological, and economically important factors should also be considered during the development of any species' management strategy (Abell, Allan, & Lehner, 2007;Suski & Cooke, 2007;Granek et al., 2008 Mongolian salmonids may be to link priority river protection across species in order to even further maximize the resources available to authorities. Therefore, the Shishged, Delgermoron, Orkhon, Tuul, and Onon rivers represent genetic hotspots as each was deemed important for two or three of the species investigated and thus these regions should be made the focus of initial conservation efforts along with Lake Hovsgol for T. baicalensis and the Kherlen River for B. lenok (Figure 8a-c). If critical habitat can be sufficiently protected, local fish densities are likely to increase with a higher number of individuals then being able to emigrate to neighboring river systems over time (Abell et al., 2007). A key recommendation for targeting these genetically valuable populations would be to establish a network of spatially protected areas to improve their overall protection and survival. Freshwater Protected Areas (FPAs) have been successfully implemented in many countries around the world to conserve genetic diversity and aid in the preservation and recovery of threatened and exploited fish populations (Abell et al., 2007;Suski & Cooke, 2007).

| CON CLUS IONS
Attaining a detailed knowledge of the genetic structure and intraspecific diversity of the main target species in Mongolia's rapidly emerging recreational fishery will help guide necessary improvements and benefit the development of more comprehensive national fisheries conservation and management strategies.
This information is particularly important due to the current widespread anthropogenic pressures that continue to impact resident fish populations. Additionally with this new understanding, future translocations or introductions of genetically dissimilar individuals can be avoided and inbreeding minimized within the remaining fragmented populations (Balakirev, Romanov, Mikheev, & Ayala, 2013;Hänfling, Durka, & Brandl, 2004;Hänfling & Weetman, 2006;McDougall, Welsh, Gosselin, Anderson, & Nelson, 2017;Slynko et al., 2015). However, in order to improve the outcomes  (Tong pers. comms., in Hogan & Jensen, 2013). Thus, even more urgent actions are needed in both of these countries to avoid further declines and growing local and regions extinction rates. The potential to effectively transfer the current Mongolian research results and recommendations to help guide such changes is highly plausible and necessary as long as intraspecific genetic variation can be determined and new management measures effectively implemented within the existing jurisdictional framework relating to regional fisheries strategies. Closer cooperation among scientific and governmental entities at all levels, a proposal that has been made recently in Europe (Weiss, Kopun, & Sušnik Bajec, 2013), could also help to advance the management of these and other highly mobile threatened species across northern Asia. Unfortunately, the current lack of political cooperation and environmental management coordination between countries, particularly in regard to the delegation of natural resources, remains an even greater barrier to forming transnational joint fisheries management plans across the region.

ACK N OWLED G M ENTS
This research was conducted as part of the IWRM in Central Asia: Model Region Mongolia (IWRM MoMo) project in its second phase

DATA ACCE SS I B I LIT Y
The authors agree to submit the complete data set obtained during the current study to GenBank upon acceptance of this manuscript in order for it to be publically available by publication.