Abstract
The taphonomic varieties of over 800 specimens of Kimberella (collected from the Vendian rocks of the White Sea region) provide new evidence of the animal's anatomy such as: shell morphology, proboscis, mantle, possibly respiratory folds and possibly musculature, stomach and glands. Feeding tracks, crawling trails and, presumably, escape structures preserved along with the body imprint provide insights on the mode of locomotion and feeding of this animal. The shield-like dorsal shell reached up to 15 cm in length, 5–7 cm in width, and 3–4 cm in height. The shell was stiff but flexible. Evidence of dorso-ventral musculature and fine transverse ventral musculature suggests arrangement in a metameric pattern. Locomotion may have been by means of peristaltic waves, both within the sediment and over the surface of the sea floor, by means of a foot resembling that of monoplacophorans. Respiration may have been through a circumpedal folded strip (possibly an extension of the mantle). Feeding was accomplished by a retractable proboscis bearing terminal hook-like organs and provided with a pair of structures interpreted here as glands. Whilst feeding, Kimberella moved backwards. The structural complexity of Kimberella poses questions about the time of origin of the triploblastic metazoans.
The Late Precambrian Kimberella quadrata Glaessner & Wade, 1966 was originally described from the late Precambrian Pound Quartzite of Ediacara Hills, South Australia. Originally considered as a problematic fossil, possibly belonging to the Siphonophora (Glaessner, in Glaessner & Daily 1959), Kimberella was then reconstructed as a medusa of uncertain affinities (Glaessner & Wade 1966). Known from a few specimens, it was later interpreted as a cnidarian pelagic medusa, closely related to extant cubozoans or box jellies (Wade 1972). Later authors have reconstructed Kimberella as an animal similar to extant chirodropid cubozoans or sea wasps (Glaessner 1984; Jenkins 1984, 1992; Gehling 1991). Thus reconstructed, Kimberella has been used as one of the best examples of a metazoan lineage crossing the Precambrian-Cambrian boundary with essentially no morphological change up to the present day. It has been cited as one of the most convincing counterexamples against hypotheses grouping all Vendian macrofossils in non-metazoan higher taxa (sensu Seilacher 1992). The acceptance of Kimberella as an essentially modern cubozoan also had major palaeoecological implications: living chirodropids are fast-swimming predators with powerful venom, and Kimberella thus has also been reconstructed by some as a coelenterate predator (Jenkins 1992).
New fossil material from the Vendian of the White Sea region forced reinterpretation of Kimberella as a mollusc-like animal (Fedonkin & Waggoner 1997) with a high dorsal, non-mineralized shell and an animal capable of active locomotion. The radioisotope U–Pb age 555.3 ± 0.3 Ma of a zircon from a volcanic ash bed in the Vendian marine siliciclastic rocks inside the stratigraphic range of Kimberella in the Winter Coast of the White Sea region was, therefore, assumed as the minimum age for the oldest triploblastic invertebrates (Martin et al. 2000). This is the oldest calibration point of molecular clocks for the Precambrian bilaterians. Co-occurrence of the body fossils and trace fossils produced by Kimberella revealed the mode of feeding and locomotion of this animal (Ivantsov & Fedonkin 2001; Fedonkin 2001, 2003; Gehling at al. 2005).
In this paper we describe a new fossil collection of unique preservation, which sheds light on the previously unknown aspects of morphology, anatomy and lifestyle of Kimberella.
Stratigraphic setting
Over 800 specimens of Kimberella have been excavated from the Vendian siliciclastic rocks in the White Sea region, northern Russia during the last few years. Major fossil localities were discovered in the deposits exposed in the valleys of rivers of the Onega Peninsula (Suzma, Karakhta, and Solza rivers) and on Zimny Bereg (Winter Coast) of the White Sea. Most of the collection described here has been systematically excavated by the field teams led by A. Ivantsov. Some specimens were found in the core of boreholes. All the fossils are preserved on the soles of the fine-grain sandstones or mudstone beds intercalated with the clay lamina within the Verkhovka, Zimnie Gory and Yorga formations of the Vendian. Stratigraphy and the sedimentology of the fossiliferous deposits have been described by Grazhdankin (2003). These sedimentary successions were formed in offshore shallow water marine conditions under influence of a vast deltaic system, which prograded towards the SW from the Kanin–Timan fold-and-thrust belt. The stratigraphic range of Kimberella exceeds the radiometrically dated time interval 555–558 Ma based on ages determined on zircons from the volcanic ash beds (Martin et al. 2000; Grazhdankin 2003).
Taphonomy and ecology
Environmental conditions of the deltaic platform of the Vendian Basin were characterized by sporadic turbid flows initiated by seasonal precipitation, snow melt or storm events. These ‘density currents’ produced deep erosion that modified bottom relief and sporadically brought about catastrophic sedimentation, burying entire benthic populations in situ. Preservation of Kimberella on the sand-clay interface is often associated with the imprints of bacterial mats (‘elephant skin structure’ and other kinds of shagreen surfaces) and abundant benthic algae such as the filamentous Striatella and the net-like Orbisiana (Grazhdankin & Ivantsov 1996). This indicates that Kimberella preferred calm, well-aerated conditions of shallow marine habitats, well inside the photic zone where benthic photosynthesis could be quite effective. These conditions seem to have existed long enough to allow epibenthic communities to form and reach maturity. This assumption can be substantiated by the usual association of Kimberella with other benthic organisms, such as the vagile bilaterians Yorgia, Dickinsonia and Andiva, discoidal Tribrachidium and frond-like Charniodiscus. The co-occurrence of individuals of different species with body sizes ranging from 0.2 to 50 cm in length in their life positions, and often associated with traces of their activity on the vast bedding planes, appears to represent actual community structure.
We conclude that the depth of Kimberella's habitats ranged from the uppermost subtidal zone down to a depth of tens of metres. That most of the fossiliferous sediments seem to be formed below the storm wave base does not necessarily reflect a great depth in a broad, shallow marine basin. The habitat would probably have provided a reasonably high primary production due to the availability of dissolved mineral nutrients eroded from a nearby subaerial landmass. This conclusion is supported by the fact that the Vendian metazoan fossil assemblages had a very high density, comparable to that of the recent populations living in shallow marine environments. Sedentary benthic species such as Nemiana and the burrowing organisms that produced the traces Skolithos, may reach 100 and 400 individuals per square decimetre. In the Solza River basin, A. Ivantsov was able to trace one fossiliferous layer whose surface was completely processed by grazing Kimberella for a few hundred metres laterally. This fossiliferous bedding plane bears both the trace fossils and numerous body imprints of small individuals (<1 cm long), perhaps reflecting initial colonization of an open habitat by newly hatched Kimberella.
The taphonomic variability of Kimberella is very high. Preservational diversity allows detailed reconstruction of the external morphology and even the internal anatomy, as well as locomotion and feeding styles. Several preservation modes of the soft body imprints of Kimberella occur in addition to the imprints of separate shells, feeding tracks, crawling trails and perhaps escape structures. This rich taphonomic spectrum for a single species of metazoan seems to have formed under influence of a variety of factors: (1) Kimberella's anatomical and morphological complexity preserved at different stages of development, from juveniles to adults; (2) a relatively high diversity of the habitat occupied by the animal; and (3) variations in the properties of the fossiliferous rocks. The case of Kimberella demonstrates the immense value of large fossil collections, systematically sampled from different localities and a variety of sedimentary facies that represent a wide spectrum of palaeoenvironments.
Most specimens of Kimberella seem to preserve the relative positions that they occupied in life on the sea bottom. These organisms were unable to move quickly enough to escape rapid burial in sand. The initial disturbance of their surroundings caused these animals to retract all soft parts under a shell, thus initially protecting them from fouling. Very few specimens preserve what appear to be soft tissues spreading beyond the shell edge. This behaviour could certainly have helped them to survive episodic turbidity currents or storm events, but did not prevent death in the case of catastrophic sedimentation. Some individuals, usually relatively small ones, are preserved at the end of their feeding tracks or crawling trails. Of particular interest are the fossil crawling trails that cross the grazing marks of the animal as it presumably tried to escape the accumulating sediment. Such fossils reflect the dynamics of benthic life of Kimberella. Most of this latter category of fossils is represented by small juveniles commonly preserved in the laminated siltstone that do not bear any features of sudden rapid sedimentation. Slow sedimentation did not kill some larger individuals, but in this circumstance some juveniles failed to escape even though they were able to move a few centimetres through the water-saturated sediment.
Many specimens show an anterior elongated structure, here interpreted as a proboscis, and a pair of basal miniature structures that may represent two pharyngeal glands. These features as well as the imprints of a dorsal surface of the shell with a peculiar ornamentation are illustrated here for the first time. In many cases the surface of the fossil retains not only the external appearance of the organisms, but also overprints of internal anatomical structures. The overprinting phenomena have been reported before in other Vendian metazoan fossils, as in the case of the diverticula of the gut of the segmented bilaterians Dickinsonia, Yorgia and others (Jenkins 1996; Dzik & Ivantsov 2002; Ivantsov et al. 2004). The body fossils of Kimberella also provide a number of examples of overprinting, particularly of the gut and musculature. The ventral surface of the soft body is often well-preserved, in spite of post mortem deformations related to the decomposition of the tissues as well as to the compaction, dewatering and other diagenetic changes of the sediment.
A critical role in the preservation of the soft-body imprints and bioturbations may have been played by the mucus secreted by Kimberella while it was alive. Animal mucus (a complex glycoproteid mixture) could act as glue that cemented the underlying mud. If we assume that it was secreted by a dying animal and that it was immediately attacked by bacteria living in the sediment, such a mucus-clayey mixture may have produced a kind of biofilm that was cohesive and elastic enough to remain intact during the post-mortal deformations. The taphonomic process that ended in the fossilization of the soft parts can be reconstructed as follows: (1) fast sedimentation covers the animal resting on the surface of the muddy bottom; (2) formation of a mucus-clayey film under the foot of animal; (3) the soft mud from below moves upward replacing the space occupied by the decomposing soft body under the shell; (4) the cohesive film on the mud surface meets resistant structures (such as internal organs) that leave imprints over the mucous film; and (5) full decomposition of the shell and its collapse. Rapid mineralization of the sediment immediately above the decaying animal due to bacterial interaction with the decay products might have played a significant role in the formation of the body fossils of Kimberella as well. This phenomenon is described by the ‘death mask’ taphonomic model that is applicable to some Ediacara organisms (Gehling 1999). The presence of a clay layer below the fossil imprint is indicative of a prolonged period of very slow sedimentation, which allowed establishment of the microbial films and bacterial mats. These were likely the feeding substrate and acted as a taphonomic factor. In addition, clay effectively decreases permeability of the sediment, thus increasing the rate of fossilization due to early diagenetic mineralization. The properties of the substrate on which Kimberella lived can be deduced to some extent based on the preservation style of the finest scratch marks, which could have been preserved only in case of cohesive sediment. Some casts of the pits in the feeding tracks may be interpreted as biofilm tears.
Kimberella is always preserved in negative hyporelief, that is by depressions of various depth on the sole of the competent bed (sandstone or mudstone). This kind of preservation puts Kimberella in a category of ‘resistant’ fossil sensu Wade (1968), implying that the original structure was comparatively firm in order to retain its shape in spite of decomposition and compaction of the sediment. However, no direct signs of biomineralization have been found. There is now evidence of pyritization, common to most specimens, related to the post mortem taphonomic processes, in particular, to the activity of the sulphate reducing bacteria. Our field observations indicate that the amount of pyrite precipitated is roughly directly proportional to the body size of Kimberella, and in reverse to permeability of the sediment.
Fossil material is of Kimberella quadrata (Glaessner & Wade 1966); Vendian sequence deposits (Ediacaran in age), White Sea Region, Russia. All figured specimens are accompanied by the museum catalogue numbers of the Vendian Collection, Laboratory of Precambrian Organisms, Paleontological Institute (PIN), Russian Academy of Sciences, Moscow. Most of the figured specimens are latex casts pulled from the actual fossils preserved as impressions into the sole of the overlying sandstone layer, negative hyporelief. For photography the latex casts are coated with ammonium chloride, which helps to reveal fine morphological detail. Pictures of the rock specimens are indicated in the captions (Figs 4, 11, 13a, 19a, b). All photographs are provided with a millimetre-scale bar.
These >800 specimens are all collected from the Ust'-Pinega Formation (Upper Vendian sequence) in the southeastern White Sea Region, Arkhangel'sk District, Russia. Major fossil localities occur between Medvezhy Creek and the Yorga River along the Winter Coast of the White Sea, about 100 km north of Arkhangelsk City, and from outcrops in the valleys of Suzma, Solza and Karakhta rivers, Summer Coast, on the northeastern Onega Peninsula.
Geographic position is indicated by the first four digits of the specimen numbers: 3993, Winter Coast, White Sea; 3992, Suzma River; 4853, Solza River; 4852, Karakhta River; all the last three localities are on the northeastern Onega Peninsula, Summer Coast, White Sea.
Morphology of Kimberella
Most of the body fossils are oval-shaped, bilaterally symmetrical imprints, with several zones arranged concentrically. The length of the fossils ranges from 2–3 mm to 150 mm. The large number of specimens available may be arranged as a growth series and allows observation of the effect of taphonomic phenomena, which is informative of both external morphology and internal anatomy. Extensive use of latex replicas in the study of these fossils helped to reveal fine detail, in particular, of the smallest individuals. It is convenient to use neutral descriptive terms for characters of the body fossils and later offer interpretation of these features.
Shell
The rounded, broader end of the fossil is here considered as the posterior (aboral) end of the organism (Fig. 1a, b). In most of the specimens, it is well-preserved. The anterior (oral) end tends to be narrower than the posterior (aboral). Termination of the oral end is rather variable and may depend on its lesser mechanical durability. The best-preserved specimens demonstrate that the anterior part was a narrow, tapering structure that resembles a hood, with a clear, bilobate fore end, present even in the juveniles (Figs 2, 5a–b, e, 7b).
(a) PIN 3993–5573. (b) PIN 3993–5575. Latex peel of bed sole. Scale in mm. Overall shape of the fossil showing oral (narrow) end and aboral (wide) end, smooth external outline of the foot (left side of both specimens) and distorted shell above.
PIN 3993-5590. Latex peel of bed sole. One of the smallest specimens still identifiable as Kimberella. Almost discoidal shape can be partially explained by the flattening in the course of decomposition.
The high, dorsal shell has an elongated oval, shield-like outline, which ranges from nearly round in many of the small specimens to a much more elongated structure in larger individuals. However, some very small individuals are often strongly elongated, and it is suggested here that these individuals represent a phase when the formation of the shell had not yet begun or was just incipient. In this state, individuals were able to contract or extend themselves to a greater extent than adults. The tallest zone of the shell is bordered by a flatter limb. The hood-like structure on the narrow anterior end is often visible in the relatively larger specimens (Fig. 5a, b) and weakly developed in the smallest (Fig. 5c). The outer surface of the dorsal side of the smaller specimens is covered with numerous round protuberances, uniformly spaced over the major part of the shell. In the smallest specimens, these protuberances are all approximately the same size (Fig. 5c, d). In larger shells, the size of the protuberances decreases towards the periphery (Fig. 5a–b, e). The nature of the protuberances is not clear although there are some possible interpretations: the protuberances might be simple outgrowths marking the shell surface or, alternatively, they may represent separate initial nodules of shell formation. In connection with the latter hypothesis, an analogy of the formation of small, separate platelets in the development of late metatrochophores of some molluscs is suggested. These platelets later fuse into larger plates (as in Chiton), or disappear in others (some adult Solenogastres such as Neomenia). Alternatively, these protuberances might be the bases for fine mineral spines, such as those in recent Patella lusitanica (Archeogastropoda, Prosobranchia).
In many small specimens the peripheral shelf of the shell bears numerous, radially oriented, short ribs that are regularly spaced (Fig. 5c, d). Both protuberances and the ribs are clearly visible, even in the smallest specimens, although fewer in number (Fig. 6a). However, in some well-preserved specimens the peripheral shelf with the ribs is covered by what appears to be soft tissue (Fig. 5a–b, e). This tissue is interpreted as an edge of the mantle by analogy with modern molluscs (for instance, many prosobranchs and pulmonates) whose mantle may spread over the outer surface of the shell.
The taphonomic varieties of the shell imprints clearly demonstrate that the shell was stiff, but thin and flexible, particularly in juveniles (Fig. 6b). In the smallest specimens, the shell is almost flat, being deformed in the course of sediment compaction and organic decomposition (Fig. 6a, c). The fact that the vast majority of specimens did not preserve a shell indicates that it was rather fragile and could be subjected to fast microbial decomposition (in case of its organic composition) or chemical dissolution (in case of biomineralization) after death. Having in mind that the Vendian palaeobasins of the Russian platform lay well beyond the carbonate belt of the planet, one can assume that the biomineralization of the shell, if any, could hardly have been considerable. A sharp longitudinal folding of some deformed shells (Figs 5b, 7a), as well as broken fragments of the shell (Fig. 7b), demonstrate that the thickness of the shell in the individuals 30–40 mm long did not exceed 0.3 mm. The shell was apparently purely organic and became more rigid with growth, explaining why the larger specimens are normally straight and seldom show lateral bending. If such bending does occur, it may be due to deformations of the non-lithified sediment in late diagenesis after full decomposition of all tissues including the shell material (Figs 3 and 8). As indicated above, the smaller individuals do demonstrate some lateral bending, which is quite rare for the larger individuals (Fig. 9a, b). This may indicate some flexibility of the shell during its early development.
PIN 3993-5084. Latex peel of bed sole. Scale in cm. Such large individuals are usually characterized by their slender outline, deep medial depression and well-preserved imprints of the soft parts. However, large fossils normally do not show any remnants of the shell.
PIN 3993-5607. Rock specimen of small individual. Regular shape of deep medial invagination with some traces of segmentation may represent the internal structure such as a simple gut (well preserved in the specimens shown below).
(a) PIN 3993-5570. (b) PIN 3993-5585. (c) PIN 3993-5609. (d) PIN 3993-5599. (e) PIN 3993-5604. Latex peels of bed sole. External surface of the shell demonstrates its characteristic knobbly relief (covered with protuberances). Anterior, hood-like bilateral elongation is well expressed in larger specimens (Fig. 5a, b) and hardly developed in very small individuals (Fig. 5c) in spite of outstanding preservation of shell dorsal surface detail. No obvious growth zonation can be seen in the shell structure.
(a) PIN 4852-94. (b) PIN 4852-93. (c) PIN 3993-5564. Latex peel of bed sole. External surface of the shell in small individuals. Mode of deformation indicates the thin and flexible shell wall.
(a) PIN 4852-265. (b) PIN 3993-5554. Latex peel of bed sole. Mode of post mortem deformation of the shell. Oral end of organism with bilobate outline is well preserved and clearly visible in Figure 7b.
PIN 3993-5136. Latex peel of bed sole. Scale in cm. In spite of heavy vertical deformation of the soft tissues of a relatively large individual, its axis remains straight. Oral (upper) end is commonly more deformed than the aboral, even in large individuals.
(a) PIN 3993-5605. (b) PIN 3993-5533. Latex peel of bed sole. Examples of lateral bending of the body in small individuals.
The cast of the internal surface of the shell can be seen in some specimens in which the shell is partially preserved (Figs 1a–b, 7b, 10a–d). The internal impression of the shell is comparatively smooth in the upper part showing, in some specimens, fine transverse striations (Figs 1b, 10a, d) that perhaps related to muscle scars of the possible dorso-ventral musculature. More prominent but no less enigmatic, structures are preserved in specimens shown in the figure 1a, b. These structures occur around the periphery of the shell, close to the shelf: distinct, short, linear imprints that are oriented radially and lie lying very close to each other (almost fused at their proximal ends). One interpretation of these structures, which resemble rather deep and regularly spaced imprints on the internal side of the shell, is that they represent the attachment points (muscle scars) of the musculature of the foot retractors, similar in position and outline to such muscle scars of the Palaeozoic Pilina, a monoplacophoran.
(a) PIN 3993-6501. (b) PIN 3993-5608. (c) PIN 3993-5574. (d) PIN 3993-5581. Latex peel of bed sole. Specimens showing imprints of some internal surface structures. Transverse ridges on the surface of the latex peels (Figs 10a, d) could reflect the position of the muscle scars and metameric arrangement of the muscle myomers, but could be due to simple deformation or other causes. Degree of flexibility of the shell can be seen in Figure 10c showing overfolding of aboral end of the shell.
The absence of any growth zonation in the shell structure suggests that the shell of Kimberella was the homologue of the periostracum of later Mollusca.
Soft parts
Although the external morphology of the soft parts is usually best preserved in large specimens, details of internal anatomy are commonly preserved in the smaller fossils. The preservation of the latter can be explained by the increased chance of overprinting of the internal structures through thin external tissues. Oval imprints of the soft body are usually preserved as concentric zones, showing different relief and modes of deformation.
Anterior end
New material shows that the anterior end of the soft body had a rather complex structure. All features do not appear in every specimen, but some clearly demonstrate that at least part of the anterior end was protrusible; although the extent of such protrusion may be only indirectly inferred from the scratches the animal was apparently capable of producing on the bacterial mat upon which it fed. Again, as clearly seen in some specimens, this anterior organ housed a pharynx provided with paired sac-like structures, which may have been pharyngeal glands (comparable with those found in many extant molluscs). As many otherwise well-preserved specimens, especially larger ones, do not show any evidence of these structures it is presumed that the protrusible organ could be retracted completely under the mantle and shell (see discussion below).
Inner zone
The inner zone (bounded by a narrow proximal ridge) is flat and smooth in non-deformed individuals (Figs 11, 12), and it must be the imprint of the ventral side of the organism. However, quite often, the inner zone shows a variety of deformations, such as numerous transverse wrinkles or lobes, and some longitudinal markings. The inner zone also reflects vertical deformation: most of the imprints have longitudinal invagination that may be deep enough to involve, in some cases, most of the inner zone (Fig. 3). This invagination may be caused both by the contraction of dorso-ventral muscles when the foot was retracted and by the encroachment of the clay following the decay of the inner organs. The depth of the medial invagination can reach 20 mm in the largest fossils. In a number of cases, particularly in the large specimens, the walls of such deep invagination join, producing a slightly curved longitudinal furrow (Fig. 13a). This phenomenon may have been caused by a side tilt of the laterally collapsed shell and specimens could give a false impression on the width of the ventral side. Some specimens show one side preserved in full while another is almost completely folded in by the collapsed invagination (Figs 11, 13b). The inner zone is interpreted here as a contracted foot.
PIN 3993-4004. Rock specimen. Scale in cm. One of the largest individuals ever found. Lower portion is cut off by a sand-fill of the syneresis crack. This strongly asymmetric fossil is interpreted as a ventral impression, with the left half well-preserved, while most of the right half appears to have disappeared into a slit due to the collapse of deep invagination during the post mortem taphonomic process inside the sediment. The left side is remarkably smooth with very fine, oblique furrows which course from the medial furrow to the ‘crenulated zone’ and cut the proximal ridge. Like transverse wrinkles, these oblique furrows seem to be related to the structures, which were close to the ventral surface. The proximal ridge is interpreted here as an outline of a completely contracted foot.
PIN 3993-4006. Latex peel of bed sole. Scale in mm. Medium-sized individual with well preserved aboral end on the right and distorted oral side on the left. Smooth external band could be an imprint of the foot. The folds (crenulae) have notably irregular size.
(a) PIN 3993-5551. (b) PIN 3993-5552. Latex peels of bed sole. Scale in cm. Rather common asymmetry of Kimberella fossils (see also Fig. 11) interpreted as brought about by deep invagination of the ventral part upward under the shell during the decomposition process and subsequent collapse of the slit while the high narrow shell tilted aside.
Proximal ridge
The circular, proximal ridge present in many specimens is here considered as the margin of a contracted and, to some extent, an invaginated foot. It can thus be compared to the appearance of the contracted foot in Placophora and Monoplacophora.
Axial structure
In small specimens, the longitudinal invagination often has a rather regular, cigar-like shape with a round rear end and a tapering anterior one (Fig. 14a). Very often, a rather regular annulation or segmentation can be observed (Fig. 14b–d). This cigar-like depression is interpreted here as an imprint of a voluminous gut, not differentiated into distinct stomach and intestine, which has a well-developed musculature arranged as segmental rings.
(a) PIN 3993-5607. (b) PIN 4853-372. (c) PIN 3993-5536. (d) PIN 3993-5583. Latex peels of bed sole. Various types of preservation of a ‘cigar-shaped’ axial structure (interpreted as a straight, voluminous gut). It is suggested that this feature can be observed only in small individuals because it could be printed through the thin layer of the ventral musculature below. Regular annulations of this structure are visible in Figure 14c and partially preserved in other specimens (Figs 14b, d)—perhaps segmentation reflective of gut musculature.
Arrow-like structure
The anterior part of the longitudinal depression ends with an arrow-like structure (Fig. 15a–d). This structure joins with what we have considered to be the imprint of the main portion of the gut, the connection is not terminal but rather shifted somewhat backward (Fig. 15e) and has a swelling near the place of connection (Fig. 15f). The arrow-like structure is interpreted as either a feeding organ, which was a retractable proboscis, or invertible pharynx that had at its proximal end a pair of lateral pharyngeal glands. In many small specimens, this structure could extend well beyond the anterior end of the shell (see Fig. 15 g–i). In most specimens, the terminal part of the proboscis is bent or retracted. In one specimen, this proboscis has a terminal structure, which resembles the imprint of a funnel-like organ provided with several uniform teeth or digits, at least seven are plainly visible (Fig. 15j).
(a) PIN 3993-5253. (b) PIN 4853-314. (c) PIN 4853-364. (d) PIN 4853-369. (e) PIN 4853-161. (f) PIN 4853-375. (g) PIN 4853-361. (h) PIN 3993-5596. (i) PIN 3933-5594. (j) PIN 3933-5565. Latex peels of bed sole. Arrow-shaped structure on the oral end of the body is interpreted here as a proboscis or extendable feeding structure bearing two lateral bag-like structures oriented oblique to the proboscis. The proboscis can extend beyond the oral end of the body. The fossils demonstrate different degrees of extension of the proboscis/feeding structure (Fig. 15i). The base of the proboscis is situated at the anterior-dorsal side of the gut and surrounded by the swelling (Fig. 15e, h) that may correspond to the massive retractive musculature of the proboscis. Lateral bag-like structures are interpreted here as paired glands.
Axial band
A number of specimens demonstrate a prominent axial band of a uniform width that connects the base of the arrow-like structure, presumed proboscis or pharynx, with the rear end of the longitudinal depression, interpreted here as a stomach. This band is even visible on the surface covered with the transverse wrinkles (Fig. 16a), but it is preserved much better on the smooth surface of the longitudinal depression (Figs 19b, 14d, 16b). The axial band may have some relation to the rear terminal wrinkles well-preserved in some specimens (Figs 15j, 16a), but any plausible interpretation can be suggested.
Lobes
A few specimens show a clear division of the inner zone into large lobes, bilaterally arranged (Fig. 17a, b, see also the specimen Fig. 1d in Fedonkin and Waggoner 1997). These lobes are interpreted here as an expression of internal metameric structure which, however, does not embrace the external morphology. Two possible interpretation of these lobes are suggested: (1) an overprint of a voluminous, somewhat segmented stomach; or (2) a reflection of myomeres of a dorsoventral muscle system. In Kimberella the dorso-ventral musculature could be used for the following functions: (1) locomotion in synergy with the transverse pedal muscles; (2) to flatten the body and so to press the body liquids into the crenula and foot and expand them when the animal was stationary: the degree of their contraction could impinge on the shape of the gut. In this case, the metameric arrangement of this musculature in Kimberella could be reflected in the fossil morphology.
(a) PIN 3993-5586. (b) PIN 3993-5542. Latex peel of bed sole. Both in medium-sized specimens and in the smallest, the regular segmentation could represent the myomeres of dorsal-ventral musculature. An alternative interpretation of the bilaterally arranged ridges in Figure 17b would be numerous lateral diverticulae of a large gut (such diverticulae are common for the straight gut of Proneomenia, Solenogastra).
Transverse wrinkles
Far more common are relatively thin, regular transverse wrinkles that cover the inner zone but do not cross the proximal ridge (Figs 15j, 16a, 18a). There are at least three kinds of such wrinkles. In some specimens they are almost invisible in the axial region and deepen towards the periphery of the inner zone (Figs 15j, 17b). A second type of finer wrinkles can be traced across the inner zone (Fig. 18b). And in some specimens the wrinkles of the second kind cross with short transverse wrinkles that occupy predominantly the axial zone (Figs 16a, 18a). One cannot exclude the possibility that some wrinkles may be the result of post mortem shrinking of the body, but it is unlikely that this can be the cause of all wrinkles, as they appear too regular in size and distribution. More likely, the transverse wrinkles are the result of the post mortem contraction of transverse pedal muscles. A similar arrangement of the uniform transverse muscle fibres is a common feature of the large molluscs. As for the lateral, slightly more widely and regularly spaced wrinkles (Figs 15j, 17b), we suggest that these are diverticulae of a voluminous gut (compare with the structure of the stomach in living solenogaster Proneomenia).
Crenulated zone
The crenulation observed on the periphery of the oval imprint of Kimberella is interpreted here as a thin, lateral extension of body (Figs 11, 13a) (probably morphologically a part of the lower surface of the mantle) at its junction with the lateral side of the animal. This structure was regularly folded when retracted under the shell. This organ is likely to have had a function of respiration and thus be considered a possible predecessor of the ctenidia. The reconstruction of this structure as a folded band, rather than a circumpedal series of lappets, is supported by the following observations: (1) the crenulae never overlap each other, as would be expected in the case of free lappets; (2) adjacent crenulae may have rather different sizes as measured parallel to axis (Fig. 13b); and (3) regular preservation of the crenulae suggests that these structures were rather thick and dense (possibly due to folding).
The number of the crenulae is more or less constant, about 34 to 36, both in small (juvenile) and in the largest individuals. The width of the crenula measured parallel to the axis of the body is about 1 mm in individuals about 15 mm long, and reaches 5–6 mm in the forms that reach 80–100 mm in length. In the well-preserved specimens the distal side of the crenulated zone is straight. The crenulae are usually regular in size and spacing, but their size and shape may vary due to deformation, and in a few specimens a crenula may be twice as broad as the next one. Some crenulae tend to fold into two smaller ones; others show just small inward indentation of the distal side.
Some recent gastropods have a foot with a multifolded thin edge, actively used for swimming, notably by certain unshelled opisthobranchs, such as Oscanius and Akera (Morton 1988). A similar morphology is characteristic of some large flatworms, for example, the freshwater Baikaloplana valida that can reach over 10 cm in length (Porfirieva 1977). The crenulated band of Kimberella, by analogy with these living flatworms and molluscs, appears to be able to create a kind of running wave. This wave might improve the respiration of the animal by ventilating the mantle space under the shell. There is no evidence of gills in Kimberella, while the large surface of the multifolded crenulated zone could effectively perform the respiration function. In support of this interpretation, both monoplacophorans and limpets are noted to have developed respiratory and/or ventilatory organs arranged in a circle around the foot, in about the same position as the crenulation of Kimberella. The running wave might have assisted the locomotion of the animal over the bottom surface (gliding rather than crawling).
The more or less regular size of the crenulae observed in the fossil Kimberella may have been caused by compact folding of the thin peripheral parts of the crenulated organ during retraction of the foot and crenulated organ under the shell. Most, if not all, fossil Kimberella show the soft parts retracted under the unfavourable conditions of burial in life position. Thus, under the shell the folds might have had Ѕ-shape character in lateral view. When expanded beyond the shell, the edge of the crenulated organ might become less folded. In the best-preserved specimens, the crenulae form a full oval all way around the mantle cavity.
Growth pattern
Comparison of complete specimens showing both the imprints of the soft body and shell indicate that the growth of Kimberella was gradual. Proportions of the smallest and largest specimens are rather similar. The presence in the collection of rather broad, elongated individuals can be variously explained: taphonomic causes, morphological variability, sexual dimorphism and even the co-existence of different species. There is no evidence of discrete size classes or that growth was limited or indeterminate. There is no evidence of metamorphosis or other radical anatomical changes within the observed size range. If such a metamorphosis occurred, it must have occurred at a stage earlier than any of those represented. Maximum size in particular fossil assemblages may be related to the time available to grow before the deadly burial by the catastrophic sedimentation. The large number of relatively small individuals preserved can be explained by a high reproductive rate and dominating juvenile mortality as well as, possibly, by a greater chance of the largest individuals escaping from the sediment after fast sedimentation.
Grazing tracks
The sets of the miniature sand casts of the linear depressions have been first documented on the lower bedding surface of Precambrian quartzite within the Ediacara Member of the Rawnsley Quartzite in South Australia (Glaessner & Wade 1966). These structures were tentatively interpreted as the imprints of spicules. Later, these fossils were reinterpreted as the scratch marks produced by the feet of unknown arthropods, which were not depicted by Jenkins (1992). These enigmatic structures were found to show some similarity to features of a large meanderform grazing trace fossil recovered from peritidal carbonates of probable Late Cambrian age in Saudi Arabia (Gehling et al. 1995). Judging from the size of the individual meander loops of this Cambrian trail (that is over 50 cm), the body size of the animal was inferred to reach a metre in length.
New fossil material recovered from the Vendian rocks of the White Sea (Ivantsov & Fedonkin 2001; Fedonkin 2003) and Ediacaran rocks of South Australia (Gehling et al. 2005) demonstrated no meandering loops associated with the scratches, but instead rather fan-shaped clusters of these ‘paired scratches’. This observation excluded interpretation of these scratches as the traces left by a radula in its classical sense. Another misconception was the idea that, in the process of grazing, the trace maker was crawling over a previously formed set of scratches and expanded the fan span. To explain why the previous scratches were not destroyed, the authors postulated the presence of a microbial mat at the sediment surface. Indeed, the sediments bound by the biofilms or bacterial mats seemed to be widespread during the Proterozoic until the time of active bioturbation at the very end of the Vendian through the Early Palaeozoic.
In the course of extensive excavations on the Winter Coast of the White Sea, large bedding plane surfaces were exposed. Particularly abundant scratch marks have been documented on the lower surface of the laminated, fine-grained sandstones underlain by clays. These sediments represent the distal part of a suspension flow fan that accumulated on a flat, muddy seafloor in relatively deep conditions of the subtidal zone below the wave base. One particular sandstone bed contains thousands of the sets of these scratches, so that it is difficult to distinguish the individual fans (Fig. 19a, b). The parallel position of each pair of the scratches demonstrates that the hard parts on the end of the feeding structure resembled hook-like teeth, similar to the only pair of teeth in some Solenogastra, such as Alexandromenia valida (Grassé 1960). The absence of other trace fossils on the same bedding plane indicate that in particular environments, Kimberella was the dominant grazer, able to regulate the growth of the benthic primary producer over a vast space.
(a) PIN 4853-334. (b) PIN 3993-4074. Rock samples, sole of the sandstone beds. Scale in cm. Grazing tracks produced by Kimberella often have a fan-like pattern (Fig. 19a) indicating the ability of feeding organ to extend for about 2 cm (Fig. 19b) and to exploit a considerable area of the bottom surface from one resting position of the animal. After grazing, Kimberella moved aborally and continued grazing over the next sector, thus not moving over already-exploited territory. The high intensity of grazing is mirrored by the overprinting of many layers of tracks that cross each other. The detailed and sharply-defined shape of the paired scratch grooves is indicative of a capacity of the sediment to preserve such tracks produced by hard parts embedded in the feeding organ. Small round bodies scattered over the grazed area (but not seen beyond it) have been interpreted as faecal pellets of the grazer. Alternative interpretation: these round structures represent the sand casts of the prey organisms pulled out by Kimberella from the muddy sediment.
The morphological features of these structures, the fan-like patterns of their arrangement and mode of their distribution in the bedding plane as well as direct association with body fossils of Kimberella allow us to interpret these fossils as the grazing structures produced by means of a rather long proboscis bearing a pair of thin hook-like organs (Ivantsov & Fedonkin 2001; Fedonkin 2003). Preservation of the body imprints at the end of the cascades of the similarly oriented fan-like sets of the scratch marks (Fig. 20a, b) indicates that during the feeding process, the animal periodically moved backward. Grazing over the large area by means of the proboscis requires some adaptations allowing sufficient stabilisation of the body on the surface. This function could be performed by a large foot that spread widely and attached the animal to the bottom surface.
(a) PIN 4853-334. (b) PIN 4853-318. Latex peels of bed sole. The specimens demonstrate the transition from the grazing phase (marked by the scratches on the right in both pictures) to the locomotion, likely brought about by the disturbance related to the sedimentation event. Traces of the final movement of Kimberella, it is suggested, were produced under the sand event bed, which entrapped and killed the organisms.
An alternative suggestion has been made by A. Ivantsov, who suggests that instead of an elongate proboscis, there could have been a rather wide feeding organ, which could spread like a fan. Equipped with numerous teeth this organ scratched a large surface area of the sea floor in one simultaneous sweep collecting food particles (algae, protozoans or meiofauna) from the surface or from the uppermost zone of the sediment. However, mechanical constraints and the absence of morphological evidence do not allow us to accept this model.
Feeding
Kimberella was a selective predator or grazer and not a deposit feeder. This conclusion is based on the form of its feeding tracks and the absence, in spite of the large number of specimens, of any evidence of associated faecal cord or large pellets that are common for the non-selective mud-eaters. Scratch marks are often associated with numerous round bodies preserved in a positive relief on the sole of the bed. These bodies up to 3 mm across might correspond to the large objects (prey?) that Kimberella was able to pull out of the sediment by its proboscis. The shallow water habitats of Kimberella and its common association with the remains of microalgae and with the structures interpreted as a bacterial mat may indicate that this organism had some preferred grazing areas (Gehling et al. 2005). One cannot exclude that Kimberella may even have penetrated to tidal flats and semi-isolated lagoons. The evidence of the giant Late Cambrian trail Climactichnites, produced by a problematic soft-bodied invertebrate crawling on tidal flats (Yochelson & Fedonkin 1993) demonstrates that metazoan colonization of the subaerial environment began much earlier than it was previously thought. Kimberella was adapted to feeding on the algal or bacterial mats and could well enter the shallowest parts of the subtidal zone and even the littoral zone. In the conditions of the very short trophic chains typical for the Vendian (Fedonkin 1987), Kimberella seems to occupy the topmost position of the trophic pyramid.
Locomotion trails
The vast majority of the trace fossils produced by Kimberella are related to its grazing activity. The mode of feeding and crawling backwards help to understand why the locomotion trails are so rare: the trails have been erased by the grazing activity. However, a few specimens demonstrate directly the ability to crawl through the sediment and escape from it after the burial (Figs 21, 22). These trails are preserved in the negative relief on the sole of thin sandstone beds. The thin layer of the sediment, that buried the animals alive, was water-saturated and this allowed them to crawl and even escape. These trails represent the roof of the tunnel and often have fine longitudinal markings produced by the miniature knobs of the outer sculpture of the shell.
PIN 4853-5, 11, 12. Latex peels of bed sole. Scale in cm. Three small (juvenile?) individuals, which have produced long burrows under a thin layer of sand in a failed attempt to escape. The burrows are preserved in the negative hyporelief on the sole of the entombing sand layer. Note, that the locomotion trail is narrower than the shell imprint.
PIN 4853-316. Latex peel of bed sole. Scale in cm. Smooth part of the locomotion trail seems to represent the successful escape of Kimberella from the incoming event sand bed.
Discussion and conclusions
Study of the large sample of Kimberella specimens recovered from the White Sea region of Russia allow us to make the following conclusions concerning the grade of organization and ecology of Kimberella as well as its taxonomic positions within the bilaterians.
The dorsal shield of Kimberella appears to correspond with the periostracum of Mollusca, especially that of the monoplacophorans (Fig. 23). The metamerism that appears on its inner surface in some specimens may well be imprints of segmented dorso-pedal muscles, comparable with the muscles in monoplacophorans and chitonids (metameric muscles occur also in the forepart of the body of some Caudofoveata). The remarkable frilled fringe preserved on most specimens seems to correspond with the fringe of ctenidia. However, when fully extended it would appear as a flat, extremely thin lamella, while the folded arrangement may have been produced when the flange was partly retracted. It is also possible that it was permanently folded when inside the shell in order to increase the respiratory surface without the need to extend much beyond the border of the shell. A true circulatory system may have been absent in Kimberella, but instead there may have been a meshwork of lacunae. It seems most probable that the extension of the foot, crenulae and proboscis was achieved by hydrostatic pressure. All molluscs have an open circulatory system so an open coelom is suggested for Kimberella. Extension and retraction of the foot, respiration fringe and proboscis, as well as obvious elongation and contraction of the body, are indicative of an extensive and complex system of lacunae filled with the coelomic fluids. Coelom, thought to have evolved originally as a hydrostatic skeleton, must have reached the molluscan level of organization before the Cambrian radiation of the phylum.
(a) Reconstruction in ventral view of Kimberella quadrata with foot partly retracted. (b) Reconstruction in dorsal-oblique view of Kimberella. (c) Diagrammatic section of the reconstruction proposed for Kimberella. (d) Ventral view of Chiton for comparison with proposed interpretation of Kimberella. (e) Semi-diagrammatic section of a living Chiton (Mollusca, Placophora) for comparison with c). (f) Dorsal view of the shell of Biopulvina (Mollusca, Monoplacophora): the traces of the insertions of the dorso-pedal muscles are clearly visible. 1, proboscis and mouth; 2, crenula; 3, extension of the mantle beyond the shell; 4, foot; 5, shell (note that in Kimberella only of the periostracum); 6, ctenidia; 7, traces of the insertion of muscles related with the buccal apparatus; 8, traces of the insertions of the dorso-pedal muscles.
The fact that the respiratory folds extended well beyond the foot and the shell may have been due to two factors: (1) as this structure had not yet developed into fringed ctenidia, the inefficient circulatory lacunae required a great respiratory surface; and (2) the absence of efficient predators did not require the protection of the respiratory surface (the shell having the function of an insertion surface for metameric muscles). The fact that the ctenidia of monoplacophorans are structurally rather different from those of typical molluscs may support the idea that ctenidia evolved in molluscs independently at least two times. The respiratory fringe of Kimberella can be considered as an evolutionary predecessor of the ctenidia, which developed later via structural differentiation of the fringe.
The presence of a true foot, possibly comparable in structure with that of monoplacophorans, gives an additional insight on the level of organisation achieved by the bilaterians prior to the Cambrian. During locomotion, that the foot was rather narrow and elongated can be deduced from the structure of the trails left by the moving animals. In the grazing phase, function of the foot may have required it to spread well beyond the shell to provide additional attachment to the substrate while the proboscis was raking the food from the uppermost layer of the microbial mat.
Not all the characters observed in the fossils receive an indisputable interpretation here. Thus, the prominent circumpodial band close to the basis of the respiratory flange may be gonads, nervous tracts or a very large blood lacuna.
The case of Kimberella demonstrates that true mineralized shells, so common for the most of the molluscs in the Phanerozoic, were preceded (and that may be common for all shell bearing phyla), by the organic dorsal structures covered by or impregnated with the microsclerites so common in later strata. True shells were likely to have developed by fusion of such microsclerites.
No specimens of Kimberella have been found undergoing any kind of asexual reproduction such as fission or budding; it seems, therefore, that it must have reproduced sexually. The wide geographic occurrence of the species indicate an effective mechanism of dispersal that must have been a planktonic larval strategy, as the structure of even the most juvenile specimens found is that of an animal of limited mobility. There is no indication of seasonal reproduction because the size of the body fossils in the majority of the habitats represents the whole range of growth.
The presence of a fundamentally mollusc-like architecture in animals belonging to a typical Ediacaran faunal assemblage leads to search for corresponding ecological niches and habitats in the late Precambrian ecosystems. There is no evidence that Kimberella was a macropredator. Evidence of predation in the Vendian is quite rare. The reconstruction of Kimberella as a micropredator feeding on the protozoans or meiofaunal metazoans living by the sediment-water interface cannot be ruled out. Feeding on algae seems less probable. Most recent marine herbivores are derived from microphages, detritivores or predators, and have a post-Paleozoic origin (Vermeij & Lindberg 2000). Besides, contrary to digestion of proteins, that of cellulose requires a more complex set of enzymes.
Some implications of the molluscan nature of Kimberella require further mention. If Kimberella is assumed to be a true mollusc, then this implies that it possessed a trochophore, a spiral segmentation and this, in turn, requires that: (1) the Trochophorata are a really monophyletic assemblage (discarding for the time being the argument by Nielsen (1995) about trochophore and pseudotrochophore) or (2) the Trochophorata are a merely polyphyletic group. If we assume (1), then almost all living phyla must have begun to differentiate well before Kimberella, long before commonly assumed.
The study of the world's richest collections of Kimberella provides strong evidence for the existence of a morphologically complex, heterotrophic, triploblastic invertebrate in late Neoproterozoic marine ecosystems. If Kimberella was a mollusc, it implies that the Protostomata and Deuterostomata lineages must have diverged early, in pre-Ediacaran times: a model which is consistent with those molecular clock models that place the origin of the metazoans further back in time (Wray et al. 1996; Bromham et al. 1998; Wang et al. 1999; Hedges et al. 2004; Pisani et al. 2004; Bromham 2006) and not just an integral part of the Cambrian ‘explosion’ (Conway Morris 1997; Aris-Brosou & Yang 2003; Peterson et al. 2004; Peterson & Butterfield 2005).
Acknowledgments
The authors are grateful to J. Gehling, E. Yochelson and B. Waggoner for the stimulating discussion and to A. Mazin, who has made most of the photographs. This study is supported by the Russian Fund for Basic Research (Grant 05-05-64825) and by the President of the Russian Federation (Grant 2899.2006.5), and Program 18 of the Russian Academy of Sciences. Fieldwork in the White Sea Region of Russia was supported by the National Geographic Society. Thanks are also due to Monash University for providing funding for MAF as a Scientist in Residence during 2002–2003 and to J. Stilwell, P. Trusler, P.Vickers-Rich and one anonymous reviewer for providing valuable comments on the manuscript. This study has been carried out as part of UNESCO IGCP Project 493.
- © The Geological Society of London 2007