High‐resolution imaging sheds new light on a multi‐tier symbiotic partnership between a “walking” solitary coral, a sipunculan, and a bivalve from East Africa

Abstract Marine symbioses are integral to the persistence of ecosystem functioning in coral reefs. Solitary corals of the species Heteropsammia cochlea and Heterocyathus aequicostatus have been observed to live in symbiosis with the sipunculan worm Aspidosiphon muelleri muelleri, which inhabits a cavity within the coral, in Zanzibar (Tanzania). The symbiosis of these photosymbiotic corals enables the coral holobiont to move, in fine to coarse unconsolidated substrata, a process termed as “walking.” This allows the coral to escape sediment cover in turbid conditions which is crucial for these light‐dependent species. An additional commensalistic symbiosis of this coral‐worm holobiont is found between the Aspidosiphon worm and the cryptoendolithic bivalve Jousseaumiella sp., which resides within the cavity of the coral skeleton. To understand the morphological alterations caused by these symbioses, interspecific relationships, with respect to the carbonate structures between these three organisms, are documented using high‐resolution imaging techniques (scanning electron microscopy and µCT scanning). Documenting multi‐layered symbioses can shed light on how morphological plasticity interacts with environmental conditions to contribute to species persistence.

In the Brazilian Abrolhos Bank, a greater abundance of species was found to favor higher nutrient levels and featured sedimentshifting capabilities (Coni et al., 2017), while in the Great Barrier Reef (GBR) of Australia, corals have been observed to shift their metabolism toward more heterotrophic lifestyles (Anthony, 2000), assumedly compensating with feeding for the lowered carbon fixation by photosynthesis (Atkinson, 2011). These turbidity-resistant corals are characterized by massive growth forms and large polyps, and they show mechanisms for sediment removal (mucus production and transport via ciliary currents, tissue pulsation) (Bongaerts et al., 2012;Lasker, 1980;Logan, 1988;Stafford-Smith, 1993). Also, they are adapted to the higher nutrient loads by exhibiting higher heterotrophic rates (Anthony, 2000(Anthony, , 2006, or by hosting lineages of Symbiodinium adapted to low light levels (Garren et al., 2006;LaJeunesse et al., 2010). In addition, corals exhibit unique ectosymbionts (e.g., trapeziid crabs), which aid in the removal of sediments (Stewart et al., 2006), as well as increase flow rates within interstices of coral skeletons (e.g., Doo et al., 2018).
Previous studies of solitary corals from inshore turbid settings describe the genus Scolymia in the family Faviidae (Coni et al., 2017), a genus that, in contrast to the free-living species, comprises attached corals. Members of this genus are known to actively remove sediment off their surface by means of combined ciliary action and mucus entanglement (Logan, 1988;Tomascik & Logan, 1990). They are less susceptible to bleaching and subsequent mortality, and appear to prefer coastal areas with high sediment and nutrient loads, where they demonstrate high levels of heterotrophy (Coni et al., 2017).
Another group in the family Fungiidae, in contrast, are free-living and show highly developed mobility, ecomorphological adaptations such as abrasion-resistant skeletons, passive and active hydromechanical adaptations such as burial avoidance, and the ability to right themselves with the aid of polyps (Hoeksema & Bongaerts, 2016;Hoeksema & Moka, 1989). Furthermore, time-lapse photography of Lobastis scutaria and Herpolitha limax demonstrates their ability to remove sediment by pulsed, polyp inflation, in addition to ciliary action and mucus entanglement (Bongaerts et al., 2012). These characteristics allow for their distribution across reef-wide environments and possibly into uncolonized areas, where they could potentially seed new reefs (Sheppard, 1981), which would be an adaptive advantage under increasing sediment loads.
In this study, we add to the growing literature documenting the symbiotic associations of two solitary stony coral species from two separate families and suborders, namely, the Dendrophylliidae Heteropsammia cochlea (Spengler, 1781)

and the Caryophylliidae
Heterocyathus aequicostatus (Milne Edwards & Haime, 1848). They are associated with the boring cryptic sipunculan worm Aspidosiphon muelleri muelleri (Diesing, 1851) and the micro-bivalve Jousseaumiella sp., which resides within the sipunculan worm´s cavities. These associations were found in the neritic inter-reef channel habitats of western Zanzibar (Tanzania). They are facultatively photosymbiotic with Symbiodiniaceae dinoflagellates (Hoeksema & Matthews, 2015), which thrive in euphotic conditions. These two coral species are known to host photosymbionts not only on their upper side but also on their underside, while light is transferred through their skeleton to the Symbiodiniaceae on their underside, thus optimizing photosymbiosis (Fine et al., 2013).
Heterocyathus aequicostatus was first mentioned in 1948 (Milne Edwards & Haime, 1848), and Heteropsammia cochlea in 1926(van der Horst, 1926, from Tanzanian waters. The symbiotic relationship between H. aequicostatus and H. cochlea with the sipunculan worm species Aspidosiphon muelleri muelleri, was first described by Bouvier (1894), who interpreted this to be a commensalistic relationship. Their symbiotic relationship with the montacutid bivalve genus Jousseaumiella sp. was initially described under the name of Jousseaumia by Bouvier (1894) from Yemen, and by Bourne (1906) from Sri Lanka. Both authors interpreted it to be commensalistic, with numerous small specimens of the bivalve embedded in the skin of the posterior part of the body of the sipunculan worm, and toward the innermost coils of the worm chamber. The coralsipunculan association has also been described by Fisk (1981Fisk ( , 1983 and Goreau and Yonge (1968) who reported occurrences at Wistari Reef (southern Great Barrier Reef) and Lizard Island (Australia), and by Hoeksema and Best (1991) from Indonesia. In the Western Indian Ocean, Feustal (1965) and Yonge (1975) found that the coral larva settles on shells of dead gastropods, which were already bored by sipunculan larvae, which previously settled on the shells. Similarly, Pichon (1974) described observations from Madagascar that implied that this co-habiting type of symbiosis was initiated by the coral planula larva settling on the micro-gastropod shell, which was already inhabited by the sipunculan. When the worm grows too large for the sheltering gastropod shell, the coral is forced to provide protection by growing around the worm. This symbiotic relationship has been described as a mutualistic one, with the worm being physically protected against predators, and the coral being transported away from being buried in sediment (George, 2012), or being stabilized in lose sediment (Fine et al., 2013). In this study, we focus on the morphological characteristics and adaptive advantages of this symbiotic relationship, which previously has not yet been described in depth.
Sampling took place from August to November 2014 during the slightly cooler, drier (SE monsoon) season when slightly stronger winds and a northward-flowing surface current dominated, and the reefal waters were relatively clear. The SE monsoon is followed by an onset of the short rains, followed by the warmer and more humid NE monsoon starting in December, characterized by southward-flowing surface currents, long rains, and slightly increased turbidity of the reefal waters.
Two areas of 150 m × 50 m were investigated by two SCUBA divers in water depth between 10 and 25 m, in order to locate the solitary corals (search and recovery method; PADI, 2003). The 20 m × 2 m belt transect method (English et al., 1997)  were collected by SCUBA divers from the seafloor sediment in Changuu for thin section analysis.

| Culturing in the aquarium facility
The six live-collected corals were stored in seawater aquaria at the Institute of Marine Sciences, Zanzibar, in order to maintain the specimens in the physical conditions closest to that at the sites of collection, before transporting them to ZMT in Bremen, Germany. Upon arrival at ZMT, the corals were placed in a seawater aquarium for three weeks for photographic documentation.

| Micro-structural analyses of the solitary corals
For structural analyses and µCT scans of the live-collected corals, two coral specimens (one H. cochlea and one H. aequicostatus, see Table 1) were fixed in 40 ml of 99.8% Ethanol. Afterward they were treated with 90% H 2 O 2 for 48 hours to remove organic mat- Light microscopy was undertaken on thin sections cut from the skeletal remains collected from the sediment with a Keyence VHX-5000 equipped with a VH-Z20R lens a VHX-5020 camera and XY-Stage VHX-S550E.

| Field observations
Solitary corals were identified in situ from water depths between 16 and 21 m below sea-level from the windward-fringing reef flanks, which extends 1.5 km off the southern side of Bawe Island.

| Live-collected specimens
The live-collected specimens thrived in the aquarium culture at ZMT, and the moving ("walking") behavior was observed (see video in Figure 2).
The calice length of the three specimens collected of H. aequicostatus ranged in their long axis from 11.7 mm to 12.5 mm ( Table 1). The color of their organic tissues ranged from pale brown to dark brown ( Figure 3). The coral skeletons were sub-circular with a slightly convex base ( Figure 4). Edges of the corallum were smooth, with a roundish and approximately 0.5 mm deep calice. The imperforate theca showed one axis growing septa along the vertical axis, which was ornamented with a spike-like and granulate texture.
The three specimens of H. cochlea showed a squat base, and a flat and oval calice where the polyp emerged. The narrowed and sigmoidal calice ranged in its long axis length from 8.7 mm to 17.7 mm.
One of the specimens of H. cochlea showed two corallites, that is, budding.
All coral specimens showed an aperture of ca. 1 mm in diam-  Bawe fine pebbles were recognized, while 10 bivalves were identified in the cavity of H. cochlea ( Figure 5). As seen under SEM, the bivalves were bilaterally symmetrical, with a heterodont hinge lying in the sagittal plane. They featured two forms of hinge teeth of different sizes, with a shorter anterior ill-defined tooth and a posterior cardinal tooth, closely curving posteriorly ( Figure 7). This bivalve is identified as Jousseaumiella spp., which has previously been observed to live exclusively in commensal associations with sipunculans residing within H. aequicostatus and H. cochlea (Bourne, 1906). Bourne (1906) identified two species, namely, J. heterocyathi, which was only found in Heterocyathus, and J. heteropsammiae, which was only found in Heteropsammia. Herein, we did not identify the bivalve down to the species level.

| Sedimentary skeletal remains
Thin

| DISCUSS ION
The majority of studies have previously described the symbiosis between the coral and sipunculan worm as being mutualistic, rather than parasitic (Fisk, 1981(Fisk, , 1983Goreau & Yonge, 1968;Hoeksema & Best, 1991;Igawa et al., 2017). Our findings, further support this interpretation on the basis of the fact that the coral tissue was not affected by the lateral pores created by the sipunculan. In addition, the effect of etching of the Aspidosiphon chamber within the coral was minor (Figure 5), implying that the coral grows around the worm, rather than the worm boring into the coral skeleton (cf. Igawa et al., 2017). This interpretation is further supported by the observation, from the sedimentary remains, that the coral grows around a gastropod shell, which already features an Aspidosiphon chamber ( Figure 8). Previous studies have found traces of the original substrate, such as a gastropod or scaphopoda shell within live specimens (Fisk, 1981(Fisk, , 1983Gill & Coates, 1977;Goreau & Yonge, 1968;Stolarski et al., 2001;Zibrowius, 1998). Since the nubbin stages from the sediment reveal gastropods as substrate (Figure 8), we speculate that dissolution, reabsorption, or remineralization of the primary carbonate structure of the gastropod shell has taken place. is the benefit to these different organisms involved. If this is a successful relationship, then why is it so rare, compared to the typical symbiotic producer-in-consumer relationship between dinoflagellate endosymbionts and reef-building coral taxa? One hypothesis is that this consumer-in-consumer-type symbiosis is less frequently encountered, because it occurs across a wide range of different environments, across varying nutrient and photic conditions, and possibly within areas of greater heterotrophic plasticity.
Heterocyathus aequicostatus and Heteropsammia cochlea belong to two different families (Caryophylliidae and Dendrophylliidae, respectively), implying that this symbiosis developed independently as an example of convergence (Hoeksema & Best, 1991). A recent phylogenetic study suggests that the sipunculan worms comprise two distinct clades, both of which are associated with both of the coral species, thus this association is not species-specific (Igawa et al., 2017). In support of coevolution, the worm morphology shows plasticity, determined by the internal structure of the coral host. As the worm lives and grows within the coiled chambers of the coral, the coral chambers grow simultaneously around it, resulting in "lodging mutualism" (Igawa et al., 2017). In another associative case, for example, the sipunculan worm also inhabits polychaete worm tubes; however, as the sipunculan worm grows and fills the tube, it no longer fits and needs to move into a larger worm tube.

While the interaction between Heteropsammia sp. and
Aspidosiphon has largely been interpreted as commensalistic, Arnaud and Thomassin (1976) mention that they found the date mussel Lithophaga lessepsiana to bore into H. cochlea (then: H. michelini) just above the Aspidosiphon chamber in a complex relationship that they interpret as parasitic as it harms the coral host.
Even though there are recent reports that suggest a harmful association with serpulid worms for the host (Hoeksema et al., 2019), the majority of symbioses between corals and worms seems to be advantageous across a range of environments, and particularly in deep-sea conditions. For example, the colonial scleractinian cold water coral Desmophyllum pertusum has been observed to live in association with the polychaete worm Eunice norvegica in the deep waters of the North East Atlantic (Mueller et al., 2013). This association was interpreted to be a mutualistic symbiosis, where the polychaete positively stimulates calcification of the coral by up to four times. In turn, the coral provides substrate and shelter but also increases fitness by improving tissue assimilation and food partitioning for the worm (Mueller et al., 2013). In addition, aquaria observations suggest that E. norvegica benefits by stealing food from the coral host, while at the same time, it cleans the coral's Associations of a similar kind have been found in the geological past. Throughout the Cenozoic, fossil corals are reported to have lived in symbiosis with a sipunculan worm (Stolarski et al., 2001).
In the Cretaceous, Heterocyathus priscus is thought to have lived in symbiosis with a possibly sipunculan worm, while the Devonian tabulate coral Pleurodictyum problematicum is interpreted to have provided protection to its associated worm Hicetes sp. (Darrell & Taylor, 1993;Gerth, 1952;Stolarski et al., 2001). The significance of these associations lies in better understanding how mutual facilitation can enhance ecosystem functioning and species persistence under changing environmental conditions.
The greatest advantage and functionality of this coral-worm association may be particularly important in deeper mesophotic, turbid, sediment-laden, soft-bottom environments, which are rather unfavorable for most photosymbiotic benthic reef-dwellers. However, these associations typically occur in shallow, relatively clear water, coarse-grained, and inter-reef environments. The worm feeds on organic matter, the removal of which is well known to promote the recycling of nutrients. We speculate that the removal of detritus by the worm from the coral tissue and from the engulfing water layer, could also play an important role, similar to those observed in other ecosymbiotic organisms (e.g., porcelain crabs). Another ecosystem function of the worm could be to stabilize the sediment by removing fine-sized organic-rich sediment particles.

This corroborates the findings presented in previous literature
where a connection between the coral-sipunculan symbiosis is more prevalent in turbid conditions. The coral and sipunculan partners benefit particularly from this symbiosis by the fact that the coral can be moved over the sediment surface by the feeding worm (Bouvier, 1894;Feustel, 1965;Fisk, 1981Fisk, , 1983Gill & Coates, 1977;Goreau & Yonge, 1968;Hoeksema & Best, 1991;Yonge, 1975). The corals' internal morphology supports a hollow skeleton, which can be dragged around by a sipunculan worm above the soft bottom, and a squat flat base that is stable enough to be anchored under high energy conditions. Fine et al. (2013) observed that the sipunculan prevents burying of the coral and can anchor the coral in the substrate under strong tidal current conditions. This seems to conform well with the observations that were made in Zanzibar. It has also been shown that Heteropsammia and Heterocyathus rely on preying on zooplankton, and the sipunculan worms move them out of sediment enabling them to feed (Mehrotra et al., 2016). Fine et al. (2013) also found that this symbiosis also occurred in the nonturbid waters of the Great Barrier Reef, Australia, where the holobionts appeared to occupy unstable coarse sand bottoms. Fine et al. (2013) hypothesized that, in the coral-derived carbonate sands, optical conditions were ideal for the dinoflagellate endosymbionts of the corals to optimize light-trapping for photosynthesis on the buried underside of the coral. Thus, apparently this symbiosis potentially allows for adaptation to one of the two, or both, turbid and unstable sandy conditions.
The coral-worm mutual symbiosis may increase the corals' resilience, particularly under changing and episodically turbid conditions, as found in the reefs and inter-reef habitats off western Zanzibar.
Here, they mainly occur in deeper waters, but not exceeding 20 m.
In these reef channels, particles were commonly in suspension in the water column, with limited settling, while substrates were characterized as relatively clean, composed of unconsolidated, medium to very coarse, poorly-sorted biogenic carbonates grains, and very little fines. At our study site, anthropogenic disturbances included overfishing (Lokrantz et al., 2009), port, and channel dredging activities and land-based (untreated sewage) pollution (Moynihan et al., 2012), most of these were related to rapid population growth and tourism (Lange & Jiddawi, 2009). Sediment fluxes were known to range from 0.2 to 41.5 mg cm 2 d −1 (Muzuka et al., 2010), and thus peak above threshold limits of many scleractinian corals, which is >10 mg cm −2 d −1 (Rogers, 1990). Obviously H. aequicostathus and H.
cochlea are morphologically adapted to withstand variable sediment fluxes, episodic low water quality (nutrient influx), and moderate to strong currents. Stolarski et al. (2001) and Zibrowius (1998)  Such a position is presumably controlled by the sipunculan that controls the openings facing the sediment-water interface where gradients in physical-chemical properties are suitable (Santschi et al., 1990). The functionality of these pores has been extensively reviewed (Fisk, 1981(Fisk, , 1983Gill & Coates, 1977;Goreau & Yonge, 1968;Stolarski et al., 2001;Zibrowius, 1998). Two main functions have been discussed, namely, the circulation of water (Feustel, 1965;Ikeda, 1922), and the release of excrements of the sipunculan (Semper, 1880;Sluiter, 1902). Other potential functions have been discussed to include the housing of unknown boring organisms (Schindewolf, 1959), or the release of nematocysts as protection measures by the sipunculan (Bourne, 1906). Stolarski et al. (2001) proposed that the origin of the polyporous morphology could be a perforation by the sipunculan using "minute asperities which beset the proboscis" (Tenison−Woods, 1880: p. 298); or that the pores were formed by the coral, when growing around extensions of the sipunculan (Jousseaume in Bouvier, 1894). Cutler (1965, however, pointed out that these appendices seemed not to exist. A third hypothesis states that the pores were pinched off from the orifice during growth (Sluiter, 1902), while a fourth one proposes that they were bored into the corallum by some other organism (Schindewolf, 1959). Chemical dissolution was excluded by Jousseaume in Bouvier (1894) and by Schindewolf (1959). Most likely, however, it seems that the coral actively overgrows the substrate, except for an efferent pore for the expandable introvert of the worm. With time, the worm continues to grow in a spiral pattern around the coral's base, and with continued growth, the coral coenosteum maintains a full cover for the worm by adaptive extra calcification (see also Beuck et al., 2007). Polyporus corolla, thus, document a periodical re-orientation of the efferent pore in a dynamic growth interaction between the host and the symbiont organism (Cairns, 2001). The ontogenetically older pores are in many cases arranged in a roughly linear pattern, and usually show reduced diameters as compared to the polyporus corolla. They are thought to functionally assist in the facilitation of fluid exchange and in respiration (Moseley, 1881).
While in the symbiosis described herein, the coral seems to grow around the sipunculan, A. muelleri muelleri generally is able to excavate its protective home in calcareous substrate by boring into dead coral skeleton, carbonate rocks, including submerged archaeological objects (Antonelli et al., 2015;Rice, 1969). This excavating behavior differs largely from mutualistic interaction between the live coral-sipunculan consortium as discussed above.
The anatomy of A. muelleri muelleri follows the characteristic sipunculan body plan ( Figure 9) comprising a thickened posterior trunk and a narrower anterior introvert that can be retracted into the trunk (Rice, 1993). It is the expandable introvert that enables the mobility of the entire consortium. The introvert itself bears the tentacle crown and the mouth. A common feature in the Aspidosiphonidae is the possession of cuticular elaborations or projections such as hooks, spines, and papillae on the introvert.
It is assumed that the hooks and spines mechanically support the bioerosion process (Rice, 1969(Rice, , 1993 by scraping-off biochemically etched crystalline bonds through chelating agents or acids (Williams & Margolis, 1974).
The third player in the coral-sipunculan consortium is the small galeommatid bivalve, assigned as Jousseaumiella, of which three species are listed by WoRMS (2021) High resolution imaging techniques provide an excellent tool for increasing our understanding of the morphological co-adaptations and beneficial traits found in mutualistic, consumer-in-consumer, and symbiotic-type associations. It is unknown how these symbiotic interactions will be further influenced by environmental changes and what affect this will have on the ecosystems in which they occur.
Therefore, future studies should take into consideration the array of organisms associated with corals, which contribute to their productivity, functionality, and protection.

ACK N OWLED G M ENTS
We thank the Tanzanian Ministry of Natural Resources and tourism for the permit to study these organisms of Zanzibar. The Heterocyathus were ex-and imported under the CITES permit number E 01240/14. We thank the University of Dar-es-Salaam (Institute of Marine Science in Zanzibar) for support and collaboration. The ZMT with its Leibniz Graduate School SUTAS supported this study.
Particular thanks go to the research aquarium facility team of ZMT.
Salome Huthfilter of the library at ZMT is thanked for her unceasing support in accessing literature. The team of the experimental ecology facility (MAREE) of ZMT is thanked for their help in keeping the coral holobionts in the aquaria. SSD was funded through the Alexander von Humboldt Foundation. Many thanks to the Chief Editor and Associate Editor and an anonymous reviewer for their comments that were very helpful to improve this paper. Open Access funding enabled and organized by Projekt DEAL.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.