Discovery of a silicate rock-boring organism and macrobioerosion in fresh water

Macrobioerosion is a common process in marine ecosystems. Many types of rock-boring organisms break down hard substrates, particularly carbonate rocks and calcareous structures such as dead corals and shells. In paleontology, the presence of rocks with boreholes and fossil macroboring assemblage members is one of the primary diagnostic features of shallow marine paleo-environments. Here we describe a silicate rock-boring organism and an associated community in submerged siltstone rock outcrops in Kaladan River, Myanmar. The rock-boring mussel Lignopholas fluminalis is a close relative of the marine piddocks, and its borings belong to the ichnospecies Gastrochaenolites anauchen. The neotectonic uplift of the area leading to gradual decrease of the sea level with subsequent shift from estuarine to freshwater environment was the most likely driver for the origin of this community. Our findings highlight that rocks with macroborings are not an exclusive indicator of marine paleo-ecosystems, but may also reflect freshwater habitats. Macrobioerosion, the boring of rock and other hard substrates by living organisms, is used as a marker of marine paleo-environments. Here, Bolotov et al. describe a rock-boring mussel and its associated community from freshwater in Myanmar, demonstrating that macrobioerosion is a wider phenomenon.

B ioerosion is a process by which a living organism incises or bores different hard substrates (e.g. rocks, shells, corals, and bones) by mechanical disruption and/or chemical decomposition 1 . The eroding activities of living organisms are important factors in marine sedimentation and benthic ecology 2 . Bioerosion increases species diversity of marine hard substrate communities by increasing habitat complexity or as a result of the increase in accessible surface area for colonization 3 . Additionally, this process influences the evolution of coastal profiles over long timescales 4,5 . With respect to their environmental impact, erosional organisms represent an important group of ecosystem engineers 6,7 .
The investigation of bioerosion patterns supplies significant background information to other research areas (e.g., zoology, paleoecology, biogeochemistry, sedimentology, and biostratigraphy) 1,8 . Macrobioerosion trace fossils are important indicators of intertidal and shallow subtidal marine paleo-environments 9-11 , e.g., boring bivalves may be used as biological sea level indicators marking ancient shorelines 2,11,12 . Boring organisms have a great economic impact because they damage ships, fishing equipment, archeological heritage, various infrastructures in marine environments, and the shells of aquaculture molluscs [12][13][14] .
Rock borers occur in a broad range of taxonomic groups, including bivalves, gastropods, polychaetes and sipunculans (Annelida), sea urchins, sponges, and others [8][9][10]15 . The piddocks, or rock-boring mussels (Mollusca: Bivalvia: Pholadidae), are predominantly marine animals specialized for boring into a plethora of available substrates such as soft silicate and carbonate rocks, clay, corals, wood, and peat 3,14,16 . The genera Lignopholas Turner, 1955 and Martesia Sowerby 1824 contain several estuarine and marine wood-boring bivalve species, including a single species that can bore into living mangrove trees 17 . There is a unique record of clavate borings and fossil specimens of Martesia sp. from Cretaceous Burmese amber 11 . However, Lignopholas fluminalis was also collected from soft argillite rocks and brickworks in brackish sections of rivers in India and Myanmar [17][18][19] . The piddocks have developed a variety of adaptations to accommodate their rock-boring behavior and make their borings mechanically, by scraping at the substratum 20 .
In the marine environment, almost all macroborings are recorded in calcareous substrates, e.g., corals, shells, and carbonate rocks 10,21 . There are far fewer examples of fossil and recent marine borings found in non-calcareous rocks and minerals such as siltstones, quartzites, basalts, andesites, dolerites, gneisses, and others 17,21,22 . From an ichnological point of view, clavate macroborings in hardgrounds and in fossil wood belong to two ichnogenera, i.e., Gastrochaenolites Leymerie 1842 and Teredolites Leymerie 1842 for borings in lithic and lignic substrates, respectively 23 . The Gastrochaenolites borings are known from the Early Ordovician to Recent and they are mainly produced by marine endolithic bivalves 8,9 . However, several recent gastropods and sipunculan worms may also produce Gastrochaenolites 9 .
In general, macrobioerosion was considered an exclusively marine and, to a lesser degree, an estuarine process 8,10 . However, we discovered the macrobioerosion of silicate siltstone rocks caused by a rock-boring mussel species in a freshwater section of the Kaladan River, Myanmar, which greatly expands our knowledge of bioerosion. Our phylogenetic and biogeographic modeling shows that at least three members of this rock-boring assemblage are relict marine-derived lineages, which emphasizes a broad-scale expansion of saltwater taxa into freshwater environment in Southeast Asia.

Results
Discovery of freshwater macrobioerosion. The site is located in the middle reaches of Kaladan River: 21.0094°N, 92.9813°E, altitude of 11 m a.s.l., Rakhine State, western Myanmar (Fig. 1a, b).
The rock-boring assemblage is associated with submerged black siltstone rocks located in the lower part and bottom of the river valley, with an above-water outcrop near the bioerosion site ( Fig. 1b-d). The submerged rocks are located within a riverine section of~500 m in length and of~300 m in width, with depth values of 1-2 m and more.
Freshwater environment. The model of tidal influence reveals that the distance between the rock-borer's site and the upper level of the Kaladan estuary is 71 km (Fig. 2). Dating of the shift from estuarine to freshwater environment at the site suggests that it was approximately 3.5-14 Kyr ago, i.e., not earlier than the Late Pleistocene (Fig. 2). The analysis of a water sample confirms that it is a freshwater section of the river, with a chloride ion concentration of 2.7 mg/L, sodium ion concentration of 11.6 mg/L, and a total salinity of 0.16‰ (Supplementary Table 1).
Ichnotaxonomy. The macroborings from the Kaladan River are clavate, circular in cross-section throughout the length, expanded gradually below the aperture, with a greatest diameter about three-fourths of the depth; bases are rounded; sides are smooth, without clear circular or spiral bioglyph; a distinguishable neck is lacking (Fig. 3f). With respect to the lithic substrate and morphological features 24 , these borings correspond well to the ichnospecies Gastrochaenolites anauchen  Rock-boring species. Freshwater rock borers heavily invaded the submerged section of siltstone rocks that have numerous piddocklike boreholes ( Fig. 3a-f). Empty borings are sometimes filled with clay ( Fig. 3c). This boring species was determined as Lignopholas fluminalis on the basis of morphological characters (Table 1, Fig. 4a and Supplementary Note 1). The COI gene sequences indicates that it is related to the marine piddock species Barnea davidi but with low similarity (81.4%), while the nuclear 18S rRNA gene sequences show a very close relationship with several other marine Barnea spp. (similarity of 98-99%) (Supplementary Table 4). The Bayesian and maximum likelihood phylogenies reveal that this species is a sister lineage of Barnea spp. The ancestral area reconstruction models suggest that the most recent common ancestor (MRCA) of the Lignopholas + Barnea clade was a marine mussel ( Fig. 5a and Table 2). The boring mechanism of L. fluminalis is not known, but their boreholes should be made mechanically as do its marine relatives 20 .
Associated nestling species. We recorded six species of nestling macroinvertebrates that appear to be associated with this rockborer's ecosystem (Table 1, Supplementary Table 4, and Supplementary Note 1). An integrative study of the species using morphological and molecular approaches indicates that they include two clams, Scaphula deltae (Arcidae) and Novaculina gangetica (Pharidae), a gastropod, Clithon cf. reticularis (Neritidae), two polychaetes, Namalycastis indica and Neanthes  meggitti (Nereididae), and a sponge, Corvospongilla ultima (Spongillidae) (Fig. 4b-g). The results of phylogenetic and biogeographic modeling reveal that both clam taxa, S. deltae and N. gangetica, most likely represent the relict marine-derived freshwater lineages (Fig. 5b, c and Table 2). N. gangetica is a close relative of the Sinonovacula + Pharella clade ( Fig. 5b and Supplementary Table 4). S. deltae clusters together with several members of the Noetiidae but not with the Arcidae (Fig. 5c, Supplementary Fig. 2, and Supplementary Table 4).

Discussion
Here we present the discovery of a rock-boring bivalve species and associated hard substrate community in fresh water. Our record is of exceptional paleontological, archeological, and environmental importance, because, to our knowledge, until now rock-borers were known only among marine and estuarine groups of macroinvertebrates [8][9][10][11]14,15 . From freshwater habitats, only a few records of carbonate microbioerosion structures have been described, i.e., micro-borings in mollusc shells associated with a group of minute shell-boring polychaetes 25,26 .
To the best of our knowledge, silicate macrobioerosion in fresh water was previously unknown.
Our findings highlight that rocks with boreholes and fossil members of rock-borer's assemblages are not necessarily an absolute indicator of shallow marine paleo-environments, but may also reflect past freshwater riverine ecosystems. Additionally, we found that the three most important members of the assemblage, i.e., L. fluminalis, S. deltae, and N. gangetica, are marine-derived, secondary freshwater taxa. This finding is in full agreement with the hypothesis of Annandale 27 , who assumed that Scaphula and Novaculina are freshwater genera of relict marine origin. In contrast, at least one member of the rockborer's assemblage, the sponge C. ultima, is an entirely freshwater species that is not known from brackish water bodies 28 . The freshwater fauna of the lower and middle reaches of the Kaladan and Lemro River systems appears to be rather a derivative of the Indian fauna (e.g., indicator taxa such as N. gangetica 29,30 , Lamellidens aff. marginalis 31,32 , and C. ultima 28 ), but the lack of comparable molecular data from India precludes any final biogeographic conclusion. Our modeling suggests that the shift from an estuarine to a freshwater environment at the rock-borer's site was a relatively recent event that occurred after the Late Pleistocene. These results support the hypothesis that a few marine or brackish species appear to have the unusual ability to invade freshwater over rapid time scales 33 . However, we show that the invasion of a rock-boring species to a freshwater environment may lead to the origin of an associated hard substrate community representing a mix of primary estuarine and freshwater taxa.
The Gastrochaenolites has been considered a marine ichnotaxon produced by endolithic bivalves 8,23 and, to a lesser degree, by gastropods and sipunculan worms 9 . Our discovery of a freshwater silicate rock-boring organism indicates that fossil macroborings in non-calcareous substrates may have been produced not only by marine invertebrates but also by freshwater lineages of endolithic bivalves. From this perspective, the discovery of fossil pholadid bivalves in Cretaceous Burmese amber may also indicate rather a freshwater paleo-environment than the proximity of resin-producing forests to brackish waters 11 . The ichnospecies G. anauchen was described from the Early Pennsylvanian deposits of the USA 24 . Our new record supports the hypothesis of Wilson and Palmer 24 that the Pennsylvanian borings were formed by endolithic bivalves, because the recent Kaladan borings are morphologically identical to those from the Paleozoic coarse grainstone cobbles. Wilson and Palmer 24 associated such smooth-sided clavate borings with lithophagid bivalves, but the borings of the pholadid species L. fluminalis are also characterized by smooth sides. The empty borings in the freshwater rock-boring site at the Kaladan River are often inhabited by nestling bivalve species, i.e., S. deltae and N. gangetica. This observation is in agreement with paleontological records that fossil bivalve associations are complicated by frequent records of nestling bivalves, e.g., mytilids and Hiatella sp. (Hiatellidae), in the vacated borings 34 .
Human-mediated biological invasions of various species outside their native ranges are a global-scale process 35 , which are accompanied by climatic niche shifts 36 and rapid evolution of dispersal ability 37 in invaders. Among the piddocks, alien populations of Martesia striata were recorded from the Mediterranean Sea, Hawaii, Islands of the British Isles, and Florida 38 . There is a possibility of further invasion of the freshwater rock-borer's lineage outside the Kaladan River, because of the Kaladan Multi-Modal Transit Transport Project that is being implemented by the Government of India and the Government of Myanmar. The project involves the development of the transit transport system to Mizoram through Myanmar, with a 158 km inland water transport segment via the Kaladan River 39 , which may facilitate a broad expansion of the rock-borer. The larval development of L. fluminalis is unknown, but it is most likely that this species has a planktonic larva as the marine pholadids, e.g., Barnea and Martesia (Supplementary Note 1). A successful invasion of this freshwater lineage into surrounding larger Asian river basins, including the Ganges, Salween, and Mekong, can lead to unpredictable consequences for native ecosystems that can be comparable to those from other invaders with planktonic larvae 40,41 . In contrast, a freshwater population of the rockboring mussel is of exceptional importance for scientific research, as it probably represents a lineage with unique physiological adaptations to survive in freshwater environments. Taking into account its possible local range and ancient origin, this lineage should be a focus of special research and conservation efforts. Our discovery of a freshwater rock-boring assemblage in the Kaladan is in agreement with numerous occurrences of marinederived groups in rivers of the Oriental Region, which have attracted the attention of scientists for many decades 27,30,[42][43][44][45] . Fossil freshwater shark assemblages were discovered from the Late Jurassic and Early Cretaceous deposits of the Khorat Plateau in Thailand 46,47 indicating that the colonization of Asian river systems by marine animals started as early as the mid-Mesozoic. Molecular studies generally support the primary marine origin of most recent marine-derived taxa in tropical fresh waters. There were at least three separate freshwater expansions by stingrays in Southeast Asia, and brackish water habitats may have played an important role for these events as evolutionary bottlenecks 48 . Two freshwater colonization events were recorded in ariid catfishes 49 and at least one event occurs in Southeast Asian freshwater pufferfishes 50 . The phylogeography of a killifish species, Aplocheilus panchax, indicates multiple freshwater colonization events during the Pleistocene 51 . Freshwater prawns of the genus Macrobrachium independently colonized fresh waters at least five times 52 and freshwater polychaetes of the genus Namanereis show two distinct invasions 53 .
In summary, Southeast Asia seems to be a long-term arena for wide-scale expansion of marine taxa into fresh water, examples of which are known within different animal families. This pattern may be associated with the presence of a plethora of huge estuarine areas, which may facilitate the adaptation of saltwater organisms into freshwater environments. The land-level change resulting in the combination of progressive tectonic uplift of the mainland 54 and a gradual decrease of the sea level 55 may also have contributed to this process through a slow rise of river catchment elevation with a subsequent decrease in the level of tidal saltwater influence in the lower sections of the rivers (Fig. 2). Such a neotectonic evolutionary model seems to be the most appropriate explanation for origin of the freshwater rock-borer's community in Kaladan River. Several more ancient freshwater colonization events are consistent with the collision of the Asian and Indian plates and the subsequent Tethyan regression, which advanced the cohesive zone shift from shoal sea waters to freshwater environments 44 .
The freshwater lineage of L. fluminalis from the Kaladan River is an example of a rock-boring freshwater organism that provides the reliable evidence for freshwater macrobioerosion in silicate rock. Phylogenetically, the rock-boring species is a close relative The freshwater rock-borer community in Myanmar includes several nestling macroinvertebrate species, including the arc clam S. deltae and the jackknife clam N. gangetica which belong to relict marine-derived genera with brackish and freshwater species. Our tectonic modeling indicates that this unusual invertebrate community most likely originated approximately 3.5-14 Kyr ago via the neotectonic uplift of the area leading to a gradual lowering of the sea level and a subsequent shift from an estuarine to a freshwater environment. This example corresponds well to the modern data on multiple and wide-scale expansions of marine taxa into fresh water in Southeast Asia which began as early as the mid-Mesozoic. We assume that such a general biogeographic pattern may be associated with the combination of the progressive tectonic uplift of the mainland and with the subsequent decrease in the level of tidal saltwater influence in the lower sections of the rivers occurring in huge estuarine areas around the mainland. This neotectonic evolutionary model predicts the gradual shift in adaptive zone from shallow marine and brackish waters to freshwater habitats.
Our findings highlight that the rocks with macroborings and fossilized members of rock-boring communities are not a direct indicator of shallow marine paleo-environments, but may also reflect freshwater habitats. A rock-borer lineage from Kaladan River adapted to life in a freshwater environment and having putative planktonic larvae represents a potential international threat in the case it manages to invade the surrounding larger Asian river basins, including the Ganges, Mekong, and Salween Rivers.

Methods
Data sampling. The samples were collected from a site in the middle reach of Kaladan River. Several blocks with numerous borings and live representatives of the rock-boring community were obtained from submerged siltstone rocks at depths between 1.0-1.5 m using a large tommy-bar. A water sample was collected with a plastic bottle, filtered (0.45 µm) and stored in the dark before analysis by atomic absorption spectroscopy and ionic chromatography. Macroinvertebrate specimens were collected using forceps and immediately preserved in 96% ethanol. Additionally, four samples of siltstone fragments with rock-boring mussels and accessory taxa were conserved in containers with 96% ethanol.
Geographic and tectonic modeling. The model of the estuarine and freshwater sections of the Kaladan River was created using ESRI ArcGIS 10 software (www.esri.com/arcgis). The topographic base of the map was created with Natural Earth Free Vector and Raster Map Data (www.naturalearthdata.com), Vector Map (VMap) Level 0 (http://earth-info.nga.mil/publications/vmap0.html), and ASTER GDEM (https://lpdaac.usgs.gov/node/1079). The water levels were obtained from a topographic map (scale 1:500,000; The General Staff of the USSR, map nos. F46-3 and F46-4). The water level at the rock-borer's site was estimated using linear interpolation. We calculated the length of brackish section of the river based on the maximum tide height that was obtained from the Joint Archive for Sea Level 56 . The land-level change rate (i.e. uplift and sea-level change rate) is needed to reconstruct neotectonic evolution of the lower course of the Kaladan River during the Late Holocene. Unfortunately, data for this part of Myanmar are extremely scarce and uneven. South of our research location, Than Tin Aung et al. 57,58 and Wang et al. 59 have attempted to obtain such rates on the basis of field surveys of uplifted sea-level indicators. Their observations pointed out that the long-term land-level change rates in this area contain significant spatial variations. For example, late Holocene average uplift rates at southwestern Cheduba Island can be as high as 3.5-5.2 mm/year, whereas the land-level change rate landward n/a n/a n/a MF958975 n/a n/a MF959022 MF958976 n/a n/a MF959023 MF958977 n/a n/a MF959024 MF958978 n/a n/a MF959025 MF958979 n/a n/a MF959026 Mollusca: Bivalvia: Arcidae Scaphula deltae (Blanford, 1867) Nestling species, from brackish to freshwater Marine 73 MF958980 n/a n/a n/a MF958981 n/a MF959007 n/a MF958982 n/a n/a n/a  along the eastern coast of Ramree Island may be less than 0.5 mm/year 59 . These two islands are located further south of the Kaladan River, but within a~500-kmlong seismic patch that includes the downstream section of the Kaladan River 60 . Closer to the Kaladan River, Than Tin Aung et al. 58 also reports similar variations just south of the town of Sittwe, where the land-level change rate changes from more than 4.5 mm/year at the seaward sites to~2 mm/year at the landward sites.
Since the Kaladan River is located near the landward sites in these studies, we conservatively applied a long-term land-level change rate R of 0.5-2 mm/year in our tectonic modeling.
Mineralogical analyses of rock substrate. The grain size analysis was carried out on an Analysette 22 MicroTec Plus Laser Particle Sizer (Fritsch GmbH-Milling and Sizing, Germany) using a rock sample dispersed in distilled water. An XRD approach was applied to estimate the mineral composition of the substrate using a DRON-3M diffractometer (Bourevestnik Inc., Russia). The BSE images were obtained on an electronic scanning microscope Jeol JSM-6480LV (Jeol Ltd., Japan) with energy-dispersive Oxford X-Max N Silicon Drift Detector and crystaldiffractive INCA Wave-500 WDS spectrometer (Oxford Instrument Ltd., UK). The chemical composition of the rock samples was determined by an XRF analysis using a wavelength-dispersive XRF spectrometer PW 2400 (Philips Analytical, the Netherlands). The microindentation hardness (Vickers test) of the rocks was measured using a PMT-3M Vickers Microhardness Tester (LOMO, Russia) with 100 g load. The tester was calibrated using NaCl crystal with 10 g load. Several fragments of the substrate were placed into briquette and were fixed with epoxy glue with subsequent polishing of the surface. Five indentations were performed on each rock fragment, and both diagonals of indentation mark were measured. A mean microhardness value was calculated based on 20 measurements.
Ichnological identification of borings. The bivalve borings were identified based on morphological patterns and the type of substrate using the appropriate ichnotaxonomic works 23,24 . Longitudinal cross-sections of the borings were used for morphological investigation.
Morphological identification of invertebrates. For the morphological study of invertebrate specimens, we used a stereomicroscope (Leica M165C, Leica Microsystems, Germany). The comparative analysis of bivalve taxa was carried out according to the shell shape, structure of the hinge, muscle attachment scars, and umbo position. A gastropod species was identified based on shell shape and marking patterns. The polychaete taxa were identified in accordance with morphology of the external body, everted pharynx, and structures of the parapodia and chaetae. Identification of the sponge species was carried out through analysis of skeleton elements using a light microscope (Olympus CX21, Olympus Corporation, Japan) and scanning electron microscope, SEM (Tescan Vega TS5130MM, Tescan Orsay Holding, Czech Republic). Spicules of each type were purified with potassium dichromate solution (2.5% CrO 3 in 50% H 2 S), washed and mounted on a slide and specimen stub according to standard methods 61 . Measurements (n = 25) were performed under light microscope using an ocular micrometer; spine size was measured under SEM; spicule size is presented as minimum-mean-maximum dimensions.
DNA extraction and molecular analyses. New sequences were obtained from 24 invertebrate specimens belonging to seven species that were collected from the Kaladan and Lemro rivers, western Myanmar (  manufacturer's protocol. For molecular analyses, we obtained partial sequences of the following markers: the mitochondrial cytochrome c oxidase subunit I gene (COI) and 16S ribosomal RNA (16S rRNA), and the nuclear 28S ribosomal RNA (28S rRNA). We were unable to obtain the 28S rRNA gene sequences from the rockboring species, and we therefore sequenced another nuclear marker, i.e., the 18S ribosomal RNA (18S rRNA). Additionally, sequencing of the 16S rRNA gene from samples of L. fluminalis and S. deltae was also unsuccessful ( Table 1). PCR primers are shown in Supplementary Table 5. We applied four markerspecific PCR algorithms as follows: (i) COI: 95°C (4 min Searching for the nearest neighbors. Each molecular sequence was checked via the basic local alignment search tool (BLAST; blast.ncbi.nlm.nih.gov) to search for the most similar sequences in NCBI GenBank. Additionally, we used the nearestneighbor search algorithm implemented in the Barcoding of Life Data System (BOLD; www.boldsystems.org) with COI gene sequences of our specimens.
Sequence alignment. To estimate the phylogenetic position of three bivalve species, i.e., L. fluminalis, S. deltae, and N. gangetica, we sampled three sequence data sets, namely, "Pholadidae & relatives" (two partitions: 18S rRNA + 28S rRNA), "Arcidae & Noetiidae" (two partitions: 28S rRNA + 18S rRNA), and "Pharidae" (five partitions: 3 codons of COI + 16S rRNA + 28S rRNA), respectively (see Supplementary Table 6 for additional sequences obtained from GenBank). Sequences of the marine clam Cavatidens omissa Iredale, 1930 (Lucinidae) were used as an outgroup (Supplementary Table 6). The number of partitions in each data set was selected in accordance with the presence of available sequences for most closely related taxa in GenBank. The multiple sequence alignment was produced for each gene in each data set separately in MEGA6 63 , with the Muscle ("Arcidae & Noetiidae" and "Pholadidae & relatives" data sets) and ClustalW ("Pharidae" data set) algorithms. The aligned sequence data sets were inspected with GBlocks v. 0.91b (Supplementary Table 7) to exclude gaps and variable sections from the alignments using a less strict set of options. Lacking sites were coded as missing positions.
Saturation analyses and congruence of phylogenetic signals. To estimate each partition in each data set for evidence of substitution saturation, we used the test of Xia et al. with DAMBE v. 5.3.108 64 . The test revealed very little saturation even under the hypothesis of an asymmetrical tree. A partition homogeneity test was calculated in PAUP* v. 4.0a150 to check the congruence of phylogenetic signals among sequence partitions 65 . This test returned the conformity of the phylogenetic signals among the partitions in the "Arcidae & Noetiidae" data set (P = 0.99). For the "Pharidae" data set, signals were also congruent (P > 0.3), with exception of those between the COI and 28S rRNA genes (P = 0.01) although there was no significant discordance between the 16S rRNA and 28S rRNA genes (P = 0.34). We concluded that the incongruence between these data partitions does not affect the model because it appears to be the effect of copious homoplasy 31 . We therefore used the combined data set (3 codons of COI + 16S rRNA + 28S rRNA) in phylogenetic analyses. In contrast, the test indicated significant incongruence between the 18S and 28S rRNA genes in the "Pholadidae & relatives" data set (P = 0.01). However, testing of a restricted data set, which includes only the members of this family and an outgroup taxon, returned no conflicts among partitions (P = 1.00). We therefore used this "Pholadidae" data set in subsequent reconstructions.
Phylogenetic analyses. Phylogenetic analyses were carried out with RAxML v. 8 66 . The evolutionary models for each partition according to the corrected Akaike Information Criterion (AICc) of MEGA6 63 that were used under Bayesian inference framework are presented in Supplementary   Table 8. The MrBayes analyses were carried out for 25,000,000 generations with the following options: two runs, four Markov chains (three cold and one heated with temperature of 0.1), and sampling every 1000th cycle; 10% of the samples were excluded as an appropriate burn-in, and the convergence of the Markov chains was inspected with Tracer v. 1.6 69 .
Ancestral area reconstructions. To estimate the ancestral area patterns, we calculated the sets of uncalibrated phylogenetic trees for each data set, i.e., "Pholadidae", "Arcidae & Noetiidae", and "Pharidae", using BEAST 2 v. 2.4.6 with a lognormal relaxed clock algorithm and the Yule speciation process as the tree prior 70 . Models were calculated at the San Diego Supercomputer Center through the CIPRES Science Gateway 68 . We applied the same options to the data sets as in the MrBayes analyses (see above). However, we selected simplified substitution models (see Supplementary  Table 8). In each case, two independent runs with 25,000,000 generations were carried out. The phylogenies were reconstructed every 1000th cycle. The primary log data sets were inspected visually with Tracer v. 1.6 for the congruence of the Markov chains and the effective sample size of parameters (ESS) 69 . The ESS values for all parameters were found as >350, and the posterior and prior distributions were congruent. For each data set, the phylogenies from two separate runs were joined with LogCombiner v. 1.8.3 (burn-in = 10%), and the maximum clade credibility tree was reconstructed with TreeAnnotator v. 1.8.3 71 .
For each sequence data set, a sample of post-burn-in binary trees (N = 45,002) and a user-specified consensus tree obtained from BEAST runs were used to calculate ancestral area patterns on the basis of three algorithms, i.e., Statistical Dispersal-Vicariance Analysis (S-DIVA), Dispersal-Extinction Cladogenesis (DEC), and Statistical Dispersal-Extinction Cladogenesis (S-DEC) implemented in RASP v. 3.2 72 . The out-group taxa were discarded from the phylogenies with RASP v. 3.2. Additionally, only one representative of each species was used in RASP analyses to avoid uncertainty of the biogeographic models. We assigned two possible ancestral areas of the in-group species, i.e., (a) marine and (ab) estuarine to freshwater. The S-DIVA scenarios were computed with two allowed areas, allowing 100 reconstructions with maximum reconstruction for final tree = 1000, and allowing extinctions. The DEC and S-DEC scenarios were reconstructed with default options and two allowed areas. Additionally, we calculated a combined model (S-DEC + DEC + S-DIVA) joining the independent scenarios inferred from three different algorithms in RASP v. 3.2.
Data availability. The sequences generated under this study are available from GenBank. Accession numbers for each specimen are presented in Table 1