The enigmatic cnidarian Martsaphyton moxi gen. et sp. nov . from the Darriwilian (Middle Ordovician) of Estonia

. On the basis of detailed SEM­EDS study and computer tomography scanning, the single specimen of the enigmatic Martsaphyton moxi from the Aseri Stage (Darriwilian; Middle Ordovician) of Estonia was identified as a member of the phylum Cnidaria and subphylum Medusozoa, with unspecified ( incertae sedis ) class placement. Our results confirm a number of similarities in the morphology and chemical composition with previously described Sphenothallus Hall 1847 and Torellella Holm 1893, also interpreted as cnidarians. All three show tubiculous morphology, lamellar skeleton structure and calcium phosphatic composition. The new species differs from the previously described members in its unique root­like appendages, widened apertural chamber and lack of a laterally thickened skeleton wall. The element analysis revealed an unexpected chemical composition – high content of silica, iron, aluminium, potassium and sodium – of the rock matrix surrounding the specimen, which suggests possible involvement of material of volcanic origin in the sediment.

by Einar Klaamann. Although not dated, the text where Luha analysed the evolution of terrestrial ecosystems and speculated that the mysterious fossil could be one of the representatives of early plants, was plausibly the same which he had presented in 1926 at the Estonian Naturalists' Society (Perlitz 1927). There he concluded that the fossil did not belong to any faunal group and, referring to the obscure structure on the tip, stated that no animal possessed such organs that could be seen on that specimen. He compared the fossil to a group of Devonian plants, which, according to the classification of that time (Taylor et al. 2009), were known as psilophytes.
Later, thanks to its 'legendary' status, the enigmatic fossil was mentioned in another local popular science journal in the context of Earth history (Klaamann & Nestor 1976) and in a geological guidebook compiled by Nestor et al. (2007). In 2013, the fossil was retrieved from the repository of the Natural History Museum of the University of Tartu to be examined by undergraduate geology students Raili Hantson and Armin Kuningas in the course of a student project.

MATERIAL AND METHODS
The fossil is catalogued at the Natural History Museum of the University of Tartu as TUG 1303. It consists of three pieces: the main part (TUG 13031; Fig. 3A-E), a fragment from the wider upper end of the fossil (TUG  The specimen was photographed with the stereo microscope system Leica S9i and studied with the Zeiss EVO MA15 scanning electron microscope (SEM) back scattered electrons detector (BSE) uncoated in low vacuum regime. The elemental analysis was performed with an Oxford XMAX 80 energy dispersive de tector system (EDS) and Aztec Energy software at the Department of Geology, University of Tartu.
The Xray computer tomography (CT) scanning was performed at the Laboratory of Industrial Computer Tomography, Institute of Technology, Estonian University of Life Sciences. The CT scanner YXLON FF35 CT was used, the scan was performed with tube voltage 100.0 kV and tube current 750.0 μA; 3000 projections were made, with the integration time of 0.03 s.

GEOLOGICAL SETTING
The original label which came with the fossil says that the specimen was found at the Martsa quarry in North Estonia, at the edge of the Baltic Klint, near the town of Toila. Orviku (1940) lists two small quarries near the Martsa village, both of which open the same beds of the Aseri Regional Stage of middle Darriwilian age, with the thickness of 2-3 m. However, the preserved documents do not reveal which of these two quarries was the exact site of discovery of the specimen. During the Ordovician, the modern territory of Estonia was part of the Baltica palaeocontinent in the temperate climate zone in the Southern Hemisphere (Torsvik & Cocks 2013). Apart from the largely terrigenous Lower Ordovician, most of the Ordovician sequence of Estonia is characterized by carbonate rocks, predominantly limestones, which were laid down in a shallow epicontinental sea.
In North Estonia, the Aseri Stage consists of bioclastic limestones (JaansoonOrviku 1927;Orviku 1940;Hints 1997), with unevenly distributed lightcoloured phos phatic (francolitic) ooids and brown goethitic ooids. Sturesson & Bauert (1994) have suggested that the source material for these ooids came from a land area northwest of Tallinn and that the ooids were formed during a transgression/deepening event.

COMPOSITIONAL ANALYSES
The EDS chemical mapping and element analysis (Figs 4,5) showed that Martsaphyton moxi gen. et sp. nov. largely consisted of calcium phosphatic lamellae, which were, in places, encrusted with thin carbonaceous material (Fig. 5). The elemental analysis of the sur rounding rock matrix revealed elevated levels of Si, K, Al, Fe, Mg and Na (in decreasing order of relative abundance) (Fig. 6). A large portion of M. moxi is hidden in the matrix, therefore only a narrow segment can be observed and studied with classical methods without isolating the fossil from the surrounding rock and damaging the completeness of the specimen. Tomography has become an effective tool in palaeontology (Sutton 2008), allowing us to examine the threedimensional structure of the entire skeleton, investigate its morphology and also see the internal structure of the rock matrix enclosing the specimen. The exterior of the entire specimen gives a hint that the matrix is composed of two types of rock -yellowish sediment surrounding the fossil most proxi mately, and greyish sediment distally. The tomograms ( Fig. 7A-C) show the embedded tubiculous skeleton of M. moxi with rootlike appendages extending from the open surface, but also reveal that the ʽback side' is less pronounced and likely of worse preservation. The tomograms also show that at the adverse side, in places, the boundary between the two types of matrix is gradational.
Diagnosis. As for the type species by monotypy.

Martsaphyton moxi new species
Figures 3-7 Etymology. Genus name after Martsa quarry, where the fossil was found, species name after Aleksander Moks, the discoverer of the specimen.
Diagnosis. Phosphatic tubiculous skeleton with oval cross section. Skeleton has a widened apertural chamber and narrow proximal part with rootlike appendages. The regularly and asymmetrically located appendages are thin and almost perpendicular to the skeleton axis. Skeleton structure lamellar.
Description. Tubiculous skeleton, about 80 mm long, with strongly widened apertural chamber. Aperture oval (25 mm × 15 mm wide), laterally asymmetrical with one side (convex side) protruding (10 mm) further than the other side. The apertural chamber is about 30 mm long.
Walls of the apertural chamber are externally smooth with poorly developed irregular growth lines. The wall of the skeleton has lamellar structure; it is thin in the apertural part and somewhat thicker in the proximal part. In cross section, the skeleton wall seems to have equal thickness all around the proximal part. The walls of the skeleton are externally slightly concave at the transition from the narrow proximal part of the skeleton to the wide apertural chamber. The narrow proximal part of the skeleton is more than three times longer than the wide apertural chamber.
The skeleton has somewhat flattened cone shape with oval cross section in the narrow proximal part, where it shows rootlike appendages. The apertural chamber is devoid of rootlike appendages. The appendages are not preserved in full length; they are connected to the main Fig. 6. Elemental composition of the boundary between the proximal matrix ('tube') around the specimen (above) and the peripheral matrix (below). A, EDS view of the boundary region between two types of matrices; B, composite image of the EDS element distribution maps; C-J, EDS element distribution maps. body via small bumplike swellings about 3 mm in dia meter. The appendages are thin, 1-2 mm in diameter, oval in cross section, distally filled with crystalline material. The appendages are almost perpendicular to the longer axis of the skeleton, though gently proximally inclined. The locations of the rootlike appendages on the proximal part of the skeleton are regular and asymmetrical, occurring consecutively at different levels, never on the same level together. Every successive appendage has usually more than 90 o different orientation (on the plane perpendicular to the skeleton axis) as compared to the preceding appendage. The distance between consecutive appendages along the skeleton's axis is 8-9 mm. The walls of appendages show externally faint and fine longitudinal striation.
Comparison. Martsaphyton moxi gen. et sp. nov. resembles most closely the tubes of Sphenothallus Hall, 1847) in its phosphatic composition and lamellar skeleton structure, but it differs in having the rootlike appendages, widened apertural chamber and in the lack of laterally thickened skeleton wall. Martsaphyton moxi was probably attached to a hard substrate via a small disclike holdfast (lost in the fossil now, but previously photographed by A. Luha; Fig. 2A). The presence of a small holdfast also makes M. moxi similar to Sphenothallus (Van Iten et al. 1992) and conulariids (Vinn et al. 2019, p. 93, fig. 4A). Another phosphatic tubiculous fossil that slightly resembles M. moxi in its skeleton morphology and lamellar microstructure is Torellella Holm, 1893 (Vinn 2006). However, Torellella has no rootlike appendages and widened apertural chamber (Vinn 2006).

DISCUSSION
Sphenothallus and other phosphatic tubular fossils such as Torellella, which likely are phylogenetically closely linked to Martsaphyton, were all hard substrate en crusters. It is possible that Martsaphyton was also attached to hard substrate with a holdfast similar to Sphenothallus (Van Iten et al. 1992) and Torellella (Vinn 2006). If this was true, then Martsaphyton was attached to hard substrate either inside a burrow or on the seafloor. The original photograph of Martsaphyton ( Fig. 2A) shows a holdfastlike structure in the proximal part of the animal attached to a possible skeleton. Unfortunately, this part of the fossil has been lost. The lateral rootlike structures of Martsaphyton are also easier to interpret as indicating a sessile life mode. However, in asexual pro pagation in modern coronate scyphozoans stolons can be formed (Adler & Jarms 2009). These stolons are attached near the stalk of the polyp (Adler & Jarms 2009) and are somewhat similar to the appendages of Martsaphyton. Stalked echinoderms can attach to the substrate with rootlike structures known as cirri (Ruppert et al. 2004, pp. 917-927). Further cirri may occur higher up the stem (Ruppert et al. 2004, pp. 917-927) similarly to the appendages in Martsaphyton. Kozłowski (1968) illus trated and described minute rootlike structures near the apex of Conularia which slightly resemble the ap pendages of Martsaphyton, but are more chaotically arranged, increasing faster in diameter and more variably developed. Alternatively, one can hypothesize that the rootlike appendages in Martsaphyton are not attachment structures, but lateral buds similar to those of Sphenothallus sica from the Devonian of Brazil . Possible lateral buds occur also in Sphenothallus from the Mississippian of Montana, USA ). However, the lateral branches in Sphenothallus are oriented towards the aperture, but in Martsaphyton they are slightly tilted towards the apex. The latter orientation does not corroborate the idea that Martsaphyton had lateral buds. On the other hand, Martsaphyton has an apical attachment disc and had no need for additional attachment structures in the form of branches.
If Martsaphyton was a sessile invertebrate growing in upright position, it was likely feeding on suspension or planktonsize organisms. Sphenothallus and Torellella have been affiliated with cnidarians Vinn 2006) and likely were predators. Both related cnidarians Sphenothallus and Torellella occur in the early Palaeozoic of Estonia (Öpik 1927;Vinn 2006;Vinn & Kirsimäe 2015). Sphenothallus is most common in the Sandbian oil shale and carbonate rocks of northern Estonia and Torellella occurs in the upper Cambrian siliciclastic rocks (Vinn 2006).
We interpret the phosphatic lamellae of Martsaphyton as an original tube structure. The varying thicknesses of the lamellae and their variable development can be explained by the partial recrystallization of the tube wall during diagenesis. The laterally changing sharpness of boundaries of lamellae suggests diagenetic alternation of the microstructure similarly to the microstructure of Sphenothallus (Vinn & Kirsimäe 2015). The diagenetically altered ultrastructure of the phylogenetically closely related Sphenothallus and Torellella is very similar to that of the homogeneous ultrastructure of lamellae in Martsaphyton. Nevertheless, Vinn (2006) described unaltered laminae of Torellella as composed of fibres oriented parallel to the longitudinal axis of the tubes. Conulariids are the other phosphatic cnidarians presumably related to Martsaphyton; their skeletons are also composed of thin lamellae (Van Iten 1991, 1992Van Iten et al. 1992). The SEM imaging of sectioned specimens of Conularia and Paraconularia has revealed that their periderm consists of extremely thin (1-3 μm), alternating organicrich and organicpoor micro lamellae (Ford et al. 2016). This is similar to the periderm of Martsaphyton that consists of thin phosphatic lamellae, which are, in places, encrusted with thin carbonaceous material. One could speculate that biomineralization systems of phylogenetically closely related invertebrates were similar and it is likely that Martsaphyton, Sphenothallus, Torellella and conulariids shared a bio mineralization system. However, while it is also possible that similar phosphatic skeletons evolved repeatedly, this issue needs a special study.
There were also other phosphatic cnidarians in the early Palaeozoic of Estonia. Conulariids are relatively common in the Upper Ordovician carbonate rocks of northern Estonia (Estonian geocollections database at https://geocollections.info; Vinn et al. 2019). In addition to conulariids, a phosphatic problematicum Palaenigma wrangeli (Schmidt 1874) has been described from the Ordovician carbonate rocks of northern Estonia. However, the phylogenetic affinities of the latter fossil have not been recently studied and its relationships to the other tubiculous phosphatic cnidarians are unresolved.
The tubular structure around M. moxi comprises material with elevated levels of certain chemical elements that are not common in carbonate rocks in Estonia, like silica, aluminium, potassium and sodium. This suggests the possible involvement of material of volcanic origin. According to Sturesson & Bauert (1994), volcanic ash layers have been recorded at the same stratigraphical level in Sweden, but no visible indications of contemporaneous volcanic material have been observed in Estonia. The reverse side of the fossil, which is more poorly preserved than the open side, has probably been dissolved/decomposed due to chemical reactions (Figs 6, 7). The tubular structure around the specimen could be interpreted as a zone of diagenetic concentration and crystallization of mobile compounds, which are derived from dissolved volcanogenic components from the surrounding carbonate sediment.

CONCLUSIONS
Martsaphyton moxi gen. et sp. nov. is here identified as a member of the phylum Cnidaria and subphylum Medusozoa, with the class position unspecified. In the chemical composition and morphology, M. moxi can be compared to two other genera of phosphatic cnidarians -Sphenothallus Hall 1847 and Torellella Holm 1893, which also show tubiculous morphology and lamellar skeleton structure. From the latter, M. moxi differs in its unique rootlike appendages, widened apertural chamber and lack of the laterally thickened skeleton wall.
The phosphatic lamellae of Martsaphyton are inter preted as an original tube structure. The laterally changing sharpness of boundaries of lamellae suggests some dia genetic alternation of the microstructure. Martsaphyton was sessile, attached to a hard substrate with a holdfast, probably also assisted by lateral rootlike structures.
The rock matrix around the specimen shows a high content of silica, iron, aluminium, potassium and sodium. This suggests the possible involvement of material of volcanic origin. M. moxi eksemplari ümbritsev kivim on ebahariliku keemilise koostisega: suurenenud on Si, Al, K ja Na sisaldus, mis võib viidata vulkaanilise päritoluga materjalile juurdekandele Aseri eal.