A rock-boring and rock-ingesting freshwater bivalve (shipworm) from the Philippines

Shipworms are a group of wood-boring and wood-feeding bivalves of extraordinary economic, ecological and historical importance. Known in the literature since the fourth century BC, shipworms are both destructive pests and critical providers of ecosystem services. All previously described shipworms are obligate wood-borers, completing all or part of their life cycle in wood and most are thought to use wood as a primary source of nutrition. Here, we report and describe a new anatomically and morphologically divergent species of shipworm that bores in carbonate limestone rather than in woody substrates and lacks adaptations associated with wood-boring and wood digestion. The species is highly unusual in that it bores by ingesting rock and is among the very few known freshwater rock-boring macrobioeroders. The calcareous burrow linings of this species resemble fossil borings normally associated with bivalve bioerosion of wood substrates (ichnospecies Teredolites longissimus) in marginal and fully marine settings. The occurrence of this newly recognized shipworm in a lithic substrate has implications for teredinid phylogeny and evolution, and interpreting palaeoenvironmental conditions based on fossil bioerosion features.


Background
Teredinids, commonly referred to as shipworms, are a group of predominantly marine, obligate wood-boring (xylotrepetic) and wood-feeding (xylotrophic) bivalves. As the primary consumers of woody materials across the world's oceans, shipworms play a critical role in liberating and cycling the recalcitrant carbon sequestered in lignocellulosic material, from shallow coastal driftwoods to deep-sea wood falls and mangrove stilt roots, seagrass rhizomes and coconut seeds [1,2]. Yet this prolific ability to degrade wood causes billions of dollars in damage to coastal structures, seagoing vessels and fishing equipment per annum, and as a result, shipworms are considered economically important pest organisms [1].
This unusual xylotrepetic and xylotrophic lifestyle has led to several unique anatomical specializations. In contrast to most bivalves, the valves are greatly reduced in size and located at the extreme anterior end of the body, covering little of the visceral mass and offering limited protection to the animal. Instead, the shell features tiny denticulated ridges across its surface, and functions as a drilling tool for burrowing in wood [3]. Upon ingestion, wood particles are typically stored and digested in the caecum, a sac-like extension of the stomach that is characteristic of the family [2]. In lieu of protective valves, the shipworm's vermiform body is covered by a calcareous tube that is secreted onto the burrow walls. This tube surrounds the animal from the posterior edge of the valves to the burrow aperture. The burrow aperture can be sealed by a pair of calcareous, paddle-or feather-like structures, known as the pallets, which are unique to Teredinidae [2,4].
All teredinids described to date are obligate wood-borers that can complete all or part of their entire life cycle within wood [2,4,5]. This includes the giant, chemoautotrophic shipworm, Kuphus polythalamius Linnaeus 1767, which has recently been shown to settle on and reach maturity in wood prior to entering the sediment [5]. Remarkably, Lozouet & Plaziat [6] reported a shipworm species, provisionally identified as Spathoteredo sp., but known by the locals as 'antingaw', living in a 'mudstone' cliff along the Abatan River, Bohol, Philippines. Here we describe the divergent morphology, habits and phylogenetic origin of this new shipworm and on this basis propose the new genus and species Lithoredo abatanica.

Material and methods (a) Specimen collection
Specimens of L. abatanica and Nausitora sp. were collected from deposits of soft limestone and driftwood, respectively, in the Abatan River (Antequera, Bohol, Philippines), at a depth of less than 2 m, as part of a Philippine Mollusk Symbiont (PMS) International Cooperative Biodiversity Group (ICBG) expedition. The specimen location map (figure 1) was produced and approximate coordinates were determined using the National Oceanographic and Atmospheric Administration Map Viewer (https://www.nauticalcharts.noaa.gov/ENCOnline/enconline. html). Salinity was measured at the specimen collection site using a refractometer at both low and high tide. Teredinid specimens were carefully extracted from their burrows using a hammer and chisel. All individuals were photographed and specimens for microscopy and taxonomy were fixed in 4% formaldehyde solution freshly prepared from paraformaldehyde, washed and dehydrated through a graded ethanol series (30%, 50% and 70%, 30 min per wash, Â2 washes) with final storage in 70% ethanol. Additional specimens for molecular phylogenetic analysis were fixed and stored in 70% ethanol. Morphological features of soft tissues and calcareous structures were imaged using the Keyence VHX-6000 digital microscope (Osaka, Japan). All specimens were initially assigned unique specimen and lot numbers ('PMS' code) as part of the PMS-ICBG field collection. Type material deposited with the Academy of Natural Sciences of Philadelphia and Ocean Genome Legacy were provided with additional institutional accession numbers. Accession details are provided in table 1.

(b) Scanning electron microscopy
Scanning electron microscopy (SEM) was used to determine finescale shell valves features for the taxonomic description and the  presence of bacteria associated with the gill. For SEM imaging of soft tissues and hard parts, samples were dissected, critical point dried using the Samdri PVT-3D critical point dryer (Tousumis, Maryland, USA), mounted on a standard aluminium SEM stub and coated with platinum to a thickness of 5 nm using the Cressington 208 HR High Resolution Sputter Coater (Cressington Scientific Instruments, Hertfordshire, UK). Samples were imaged on the Hitachi S-4800 field emission scanning electron microscope. Gill tissues were manually fractured using a razor blade prior to coating to expose bacterial symbionts.

(c) Micro-computed tomography
Micro-computed tomography was used to visualize internal anatomy and three-dimensional spatial arrangement of soft tissues for the taxonomic description. Specimen PMS-4314 K was stained for 20 days in 10% iodine prior to imaging. Imaging was performed using a SkyScan 1173 MicroCT scanner (Bruker MicroCT, Kontich, Belgium) equipped with a Hamamatsu 130/ 300 tungsten X-ray source and a FlatPanel Sensor camera detector with 2240 Â 2240 pixels. Scanning parameters were as follows: source voltage ¼ 60 kV; source current ¼ 100 mA; exposure time ¼ 900 ms; frames averaged ¼ 3; frames acquired over 1808 ¼ 960; filter ¼ no; binning ¼ no; flat field correction ¼ activated; scanning time ¼ (x3) 00:28:14; number of connected scans ¼ 3; and, random movement ¼ ON (10). Reconstruction of the raw data was accomplished using the software provided with the scanner (NRECON 1.6.6.0, Bruker MicroCT, Kontich, Belgium). The following settings were employed to enhance image contrast and to compensate for ring and streak artefacts: smoothing ¼ no, ring artefact correction ¼ 4 and beam hardening correction ¼ activated. Reconstructed scans were then analysed using CTVox (Bruker MicroCT, Kontich, Belgium).
(d) Energy dispersive X-ray elemental analysis and X-ray diffraction mineralogical analysis Elemental analysis was performed to determine the mineralogical composition of both bored and ingested rock. Transverse sections were cut through the visceral mass of three animals (lot PMS-4174P) using a fine razor blade and analysed using the Hitachi S-4800 field emission scanning electron microscope. Sections were scanned at a working distance of 14.8 mm and an accelerating voltage of 15 kV on the secondary electron detection using the EDX Element EDS Analysis System and accompanying Element EDS Analysis Software Suite. For X-ray diffraction analyses, intestinal tissue from three animals (lot PMS-4162M) and rock samples (PMS-4315 L) were ground to an even grain size in an agate pestle and mortar in 90% ethanol, spread across a microscope slide and allowed to air-dry for 24 h. Samples were analysed on the Rigaku Miniflex 600 X-ray Diffractometer (Tokyo, Japan) with a Cu X-ray tube using the following parameters; 40 kV, 15 mA, start degree 228, stop degree 508, interval step 0.0058, speed 0.58 s 21 . Mineralogical profiles were then identified using PDXL integrated X-ray powder diffraction software version 2.8 (Tokyo, Japan).

(f ) Phylogenetic analysis
As the 18S and 28S rRNA gene sequences from both specimens (lot PMS-4316M) were identical, sequence data from only one specimen was concatenated and aligned as in Distel et al. [9] and Shipway et al. [7] with sequences representing 11 of 17 recognized genera of Teredinidae [2,5,7,10,11] [12] with the GTR þ I þ G nucleotide substitution model as selected using JMODELTEST v. 2.1.10 [13] with Akaike Information criterion. Placopecten magellanicus was specified as the outgroup, the chain length was set to 5 million, subsampling every 2000 generations and discarding the first 20% of the results as burn-in.

Results
Multiple specimens of L. abatanica were collected from carbonate limestone exposed in the bed and low banks of the Abatan River, Antequera, Bohol, Philippines (figures 1 and 2). This location is approximately 2 km upstream from station M39 of Lozouet & Plaziat [6], the only other confirmed location for this species (figure 1). Water samples taken at the site measured less than 0.5 ppt salinity at both low and high tides. Specimens ranged from 5.5 to 105.4 mm in total body length (measurements taken from living animals prior to fixation; figure 3a-c). Numerous nestling macroinvertebrate taxa were recorded occupying vacated burrow galleries of L. abatanica, including crabs, shrimp, limpets, snails (Neritidae), mussels (Mytilidae) and polychaetes (electronic supplementary material, figure S1). Although teredinids royalsocietypublishing.org/journal/rspb Proc. R. Soc. B 286: 20190434 identified as members of the genus Nausitora were found inhabiting submerged wood in the surrounding area, no teredinids matching the characteristics of those found in carbonate limestone were present in any wood examined.

(c) Phylogeny
Bayesian analysis of concatenated small (18S) and large (28S) nuclear rRNA gene sequences places L. abatanica within the Teredinidae, in a well-supported clade containing members of the genera Dicyathifer, Tamilokus and Teredora, but separated from these taxa by a branch length comparable to those separating most accepted teredinid genera. The phylogenetic position of L. abatanica among the Teredinidae is shown in figure 5 and electronic supplementary material, figure S4. Diagnosis: Pallets pentagonal with deep 'thumbnail'-like depression on outer surface and slight medial ridge on inner surface (figure 3d,e); valves lack fine-scale denticulation, but feature large, undulating ridges of spatulate denticles across the outer face (figure 3f -i); valves are anteroposteriorly narrow, and the height is approximately two times longer than width; additionally, valves are almost entirely composed of the anterior slope, with a slight posterior slope in adults and slightly more pronounced posterior slope in juveniles; caecum indistinct or absent (figure 4a); mantle thick and white to Etymology: Lithoredo, feminine, a combination of litho-(rock) and the last two syllables of the Latin noun teredo (shipworm).

Remarks:
Lithoredo is most similar to Teredora and Uperotus in that it possesses a gill that extends the entire length of the animal, from the base of the siphons to the mouth (figure 4a). However, Lithoredo is easily distinguished from these genera based on pallet and valve morphology (described below) and the absence of a caecum. A comparison of characters between Lithoredo, Tamilokus, Teredora and Uperotusphylogenetically the most closely related genera-is provided in electronic supplementary material, table S1. Lithoredo gen. nov may be distinguished from other genera within the family Teredinidae based on the calcareous structures. The valves of Lithoredo are unique among the family, with broad, undulating ridges of spatulate denticles (figure 3f-i) replacing the small acute denticles found in all currently described genera except Kuphus, where denticles are large, blunt and featureless. The simple, solid, paddle-shaped pallets of Lithoredo may be distinguished from the pallets of Bankia, Nausitora, Nototeredo Bartsch 1923, and Spathoteredo by the lack of segmentation; from Lyrodus, Teredo, Zachsia Bulatoff & Rjabtschikoff 1933, and Nivanteredo Velásquez & Shipway 2018 by the absence of a periostracum; from Teredothyra by the absence of a divided cup; from Bactronophorus Tapparone Canefri 1877 by the absence of a dagger-like extension; from Dicyathifer and Kuphus by the absence of a distinct medial ridge on the outer face; from Neoteredo by the narrow blade and slight depression; from Uperotus Guettard 1770 by the smooth surface of the thumbnail-like depression on the outer of the pallet, lacking concentric or radiating ridges; and, from both Psiloteredo Bartsch 1922, and Teredora, the pallets of which feature rounded, paddle-shaped blades compared with the angular, pentagonal profile of the blade in Lithoredo.  fig. 14-16; and pl. 10, fig. 1-3.   Remarks: L. abatanica is superficially similar to the illustration of Teredo ancilla in Barnard (1964) [14], but differs in several important ways: (1) the siphons of T. ancilla are separate along their entire length and feature few inconspicuous papillae, conversely the siphons of L. abatanica are almost entirely united (up to three-fourth of their entire length) and feature multiple rows of large papillae on both the incurrent and excurrent aperture; (2) the intestine is visible through the thin mantle on the ventral surface in T. ancilla, whereas the ventral mantle in L. abatanica is thick and opaque, and the intestine is only visible dorsally; and (3) T. ancilla was found in mangrove wood, not rock.

Discussion
Herein, we present the description of a new genus and species of teredinid from an oligohaline-freshwater river in the Philippines. Remarkable for burrowing into carbonate limestone as opposed to woody substrates ( figure 2ad), L. abatanica is anatomically and morphologically distinct from all extant taxa and raises the total number of genera in this ecologically, economically and historically important family to 18.
Several taxonomic characters identify L. abatanica as a new genus and species, including calcareous pallets distinct from those of described genera (figure 3d,e), unique valves featuring undulating ridges across the boring surface ( figure 3f -i), the absence of a caecum and the extension of the intestine into the mid-point of the anal canal (figure 4a). In addition to these unique anatomical and morphological features, phylogenetic analysis based on concatenated small (18S) and large (28S) nuclear rRNA gene sequences reveal wellsupported relationships between Lithoredo and several basal taxa within the Teredinidae (figure 5). Lithoredo is sister to Teredora, one of only two described genera (Uperotus being the other) also featuring gills that extend the entire length of the animal, from the base of the siphons to the mouth [2].
All previously described shipworm genera are obligate wood-borers [2,3], including the giant chemoautotrophic shipworm Kuphus polythalamius, which has been shown to inhabit wood early in its life cycle and is now thought to transition to sediments after initial settlement and metamorphosis on wood [5]. However, it seems unlikely that L. abatanica makes a similar transition from wood to rock. First, careful examination of submerged wood found in close proximity to bored carbonate limestone contained no evidence of L. abatanica, although this wood contained numerous specimens of a wood-boring Nausitora. Second, specimens of L. abatanica less than 5 mm in length were found burrowing in stone, suggesting that they were recently metamorphosed. Thirdly, anatomical and morphological adaptations that facilitate wood-boring and wood-digestion, including fine shell dentition and a wood-storing caecum, are absent in both large and small specimens of L. abatanica. Conversely, L. abatanica is well adapted for rock-boring, as evidenced by the thick valves with deep, excavating ridges and the large posterior adductor muscle which attaches to the posterior slope of the valves (electronic supplementary material, figure S2). Finally, it is hard to imagine a physical mechanism by which L. abatanica might transition from wood to rock, similar to K. polythalamius transitioning from wood to sediments [5]. Soft sediments can conform to the surfaces of submerged wood, allowing K. polythalamius to transition from wood to sediments without disturbing the continuity of the burrow and burrow lining. However, this is not possible for the hard rock surfaces in which L. abatanica burrows. Thus, any transition between wood and stone would be restricted to small infrequent contact patches between wood and stone. This mechanism is not consistent with the dense pattern of burrows observed over the rock surfaces and sparsity of wood observed at the collection site.
Analysis of the gut content of L. abatanica demonstrates that the ingested material has a mineral content closely matching the composition of the rock in which the animals burrow (electronic supplementary material, figure S6) indicating that this species ingests rock as it burrows. Although many other invertebrate species are known to burrow in stone [15], we are not aware of other species that burrow in stone by ingesting the substrate. We suggest that this unusual habit is a consequence of this species having evolved from a wood-feeding ancestor, since the mechanism by which shipworms burrow in wood involves both ingestion and digestion of the excavated wood [2,16]. However, it remains an open question whether L. abatanica derives any degree of sustenance from the limestone or whether the ingested rock may contribute to nutrition in other ways. For example, the ingested rock might contribute to trituration of ingested materials as do the gizzard stones of lithophagous birds and reptiles [17,18]. The observed dark colour of the gut content (e.g. figure 3b) of L. abatanica contrasts with the light-coloured rock and suggests that in addition to mineral matter, these animals may also ingest pigmented particulate material from the water column, such as planktonic algae, bacteria or terrestrial plant matter. Indeed, the large size of the gills and labial palps are consistent with a capacity for planktotrophy [2,4]. Alternatively, L. abatanica may feed on deposited materials such as green algae (e.g. figure 2a,b), cyanobacteria and other microorganisms that typically amass on and within freshwater carbonate limestone [19]. Finally, like other members of the family Teredinidae, L. abatanica may utilize a symbiont-dependent mode of nutrition. Indeed, scanning electron micrographs reveal that the gills of L. abatanica are densely populated by bacterial symbionts (electronic supplementary material, figure S5), however, it remains to be determined whether these bacteria contribute to the nutrition of L. abatanica via xylotrophic or chemoautotrophic endosymbioses as observed in other shipworms [20,21]. These questions are outside the scope of the current work but are the subject of an ongoing investigation.
The burrowing habits of L. abatanica may also have important implications with respect to bioerosion. Previously, macrobioerosion has been considered a process exclusively limited to coastal marine to estuarine environments and has only recently been documented in freshwater systems, such as the rock-boring mussel Lignopholas fluminalis (Blanford, 1867) [21 -23] and chironomid and trichopteran insect larvae [22]. Although well studied in marine and estuarine realms, freshwater settings have received limited attention and the potential impacts of freshwater organisms on the erosion of cohesive substrates has been underappreciated [23]. Our findings of a substantial population of shipworms burrowing into the limestone bedrock of the Abatan River, Philippines, greatly expand knowledge of a key ecological process in a previously poorly studied environment.
The burrowing habit of L. abatanica may also play a role in shaping its ecosystem and creating new habitats. Shipworms royalsocietypublishing.org/journal/rspb Proc. R. Soc. B 286: 20190434 are notorious for their ability to overwhelmingly colonize and rapidly degrade lignocellulosic materials [24]. In mangrove ecosystems, these animals account for approximately 70% of wood turnover [25], and the labyrinth of excavated tunnels provide refuge for fish and numerous marine invertebrates [26,27]. In this regard, the ecological impact of L. abatanica is consistent with that of other teredinids, as evidenced by the significant colonization of bedrock ( figure 2a-d), and the high fragmentation of this calcitic material strewn across the bank of the Abatan River. Additionally, several invertebrate taxa were found residing within the intricate network of derelict burrows (electronic supplementary material, figure S1). In this system, we suggest L. abatanica is the dominant ecosystem engineer; the substantial bioeroding behaviour drives niche creation, increases habitat complexity for a range of species assemblages and likely alters the course of the Abatan River.
Teredinid bivalves are largely responsible for elongate borings found in fossil marine and marginal marine woodgrounds (i.e. the ichnospecies Teredolites longissimus). Notably, where mechanical and biodegradational processes have destroyed evidence of the original wood, Teredolites-like casts and/or linings often provide the only evidence for the pre-existence of marine woodgrounds [28]. Our finding highlights that such teredinidgenerated Teredolites-like ichnofossils are not exclusive indicators of marine palaeo-woodground ecosystems but may also reflect freshwater riverine ecosystems and non-xylic habitats (figure 2d-f ). Additionally, rocks featuring extensive boreholes were considered one of the primary diagnostic features of shallow marine palaeoenvironments. Lithoredo abatanica, along with the recent discovery of the rock-boring mussel Lignopholas in Kaladan River, Myanmar [22], provides further evidence that rocks featuring macroborings are found in freshwater habitats and cannot be considered absolute indicators of marine palaeo-ecosystems ( figure 2a-d,f ).
Finally, we note that L. abatanica may be a species of conservation concern deserving of special consideration for habitat protection. Currently, the known distribution range is limited to a very restricted area along the Abatan River, Bohol, Philippines, bounded by the site described here and station M39 of Lozouet & Plaziat [6], located approximately 2 km downstream (figure 1). The observed salinities range from marginally brackish (salinity 1-2 ppt) at station M39 to freshwater less than 0.5 ppt at the location reported here, suggesting a narrow salinity tolerance of oligohalinefreshwater for this species. Furthermore, the unusual life habits of L. abatanica suggest a limited capacity for distribution to new locations. As the trophic strategy of this species remains obscure and possibly unique, and no anatomically, morphologically or ecologically similar species are known, this species should be a focus of special research and conservation efforts.

Conclusion
Herein, we describe L. abatanica, a new genus and species of shipworm, found in carbonate limestone bedrock in an oligohaline-freshwater river in the Philippines. We demonstrate that L. abatanica, in contrast to previously described shipworm species, is probably not an obligate wood-borer and lacks the anatomical and morphological specializations typically associated with wood-boring and wood-digestion in other taxa. Further, we show that L. abatanica burrows into and ingests limestone, which accumulates in the guts of animals and is expelled from the siphons as fine-grained particles. This strategy of burrowing into rock by ingestion is, to our knowledge, unique among the animal kingdom.
These findings extend and reinforce the recent conclusion that rock-boring macrobioerosion is not restricted to marine environments, and highlights the significance of L. abatanica as an ecosystem engineer, driving niche creation and increasing habitat complexity for a host of oligohaline-freshwater organisms. In addition, we suggest these findings are of significant importance to palaeontology, and demonstrate that fossils attributed to teredinids are not exclusive indicators of marine palaeo-woodground ecosystems and may also reflect freshwater riverine ecosystems and non-xylic habitats.
Ethics. All Philippine specimens used in this study were obtained using Gratuitous Permit 0140-17 issued by DA-BFAR.
Data accessibility. The datasets supporting this article have been uploaded as part of the electronic supplementary material S3.