Symbiotic fouling of Vetulicola, an early Cambrian nektonic animal

Here, we report the earliest fossil record to our knowledge of surface fouling by aggregates of small vermiform, encrusting and annulated tubular organisms associated with a mobile, nektonic host, the enigmatic Cambrian animal Vetulicola. Our material is from the exceptionally preserved early Cambrian (Epoch 2, Age 3), Chengjiang biota of Yunnan Province, southwest China, a circa 518 million-year old marine deposit. Our data show that symbiotic fouling relationships between species formed a component of the diversification of animal-rich ecosystems near the beginning of the Phanerozoic Eon, suggesting an early escalation of intimate ecologies as part of the Cambrian animal radiation.

T he early Cambrian was a time of major evolutionary change with the development of complex animal-diverse marine ecosystems. Biotic interactions are considered to have played a major role in driving taxonomic diversification and ecological escalation through this interval, although direct fossil evidence is rare. The coevolution of predators and prey has received particular attention, with fossil support from gut contents and faecal pellets, and indirectly, from adaptations for prey capture or defence from predators 1,2 . In contrast, symbiotic interactions, such as mutualism and parasitism, are difficult to diagnose from fossils, and are extremely rare in the Cambrian 3,4 .
Here, we identify a new symbiotic component to Cambrian marine ecologies, in which worm-like animals are preserved attached to a mobile host, the enigmatic animal Vetulicola. Furthermore, the attaching organisms are almost all preserved inside the mouldic vetulicolian fossils, indicating that they were attached to inner surfaces. Hosts show no sign of decay and disarticulation, host and symbiont are similarly preserved and symbionts are not found anywhere else, indicating that the symbiosis was very specific and occurred in life. The robust tubular forms of the inhabitants suggest a sedentary surface-encrusting habit rather than invasion of the soft tissues. The number of inhabitants in some specimens (>45) may have induced a negative effect on the host, the attaching endosymbionts partly obstructing water flow, i.e., biofouling of internal body surfaces.

Systematic palaeontology. Clade Bilateria, Clade Protostomia
Vermilituus gregarius gen. et sp. nov. Etymology: Genus name from vermis (Latin) meaning worm and lituus (Latin) meaning a curved trumpet, alluding to the shape of the fossils. Species name from gregarius (Latin), meaning flock or herd.
Referred material: About 192 specimens from Ercaicun, 75 from Mafang and 10 from Jianshan associated with Vetulicola rectangulata, all in the collections of the Yunnan Key Laboratory for Palaeobiology (YKLP). In total, 17 specimens from Xiaolantian and 55 specimens from Heimadi associated with Vetulicola cuneata, all in the collections of the Chengjiang Fossil Museum (CJHMD, Supplementary data file).
Locality: Ercaicun (type locality), Mafang and Jianshan localities in the Haikou area of Kunming, and Xiaolantian and Heimadi in Chengjiang County, Yunnan Province, China (for localities see ref. 5
Diagnosis for genus (monotypic) and species. Small (0.8-7.2mm long) elongated, conical tubes having three general forms, as a U-shape, J-shape or complex sinusoid, the latter being the dominant type: occasionally the tube also begins with a 360°p lanispiral coil before straightening. Coiling can be both dextral and sinistral and is in a single plane. The proximal end of the tube blunts (no bulb-like origin). Tubes increase in diameter very slowly, the proximal diameter being about 0.2 mm and the distal diameter reaching 1 mm. No longitudinal ornament. The transverse ornament of the tube consists of distinct annulation, there being about 12-16 annulae per mm. Most tubes are discrete, but in some cases two or more tubes cross. The tube wall appears to be very thin, and there is no evidence of internal septae, pseudopunctae or punctae. Paired crescentic structures are preserved at the open end of the tube in some specimens.
Host-symbiont association. All specimens of Vermilituus gregarius are associated with vetulicolians, a group of extinct animals of disputed phylogenetic affinity that possessed a convex anterior part with frontal and lateral openings, articulating with a tail-like posterior extension (Figs. 2-6; Supplementary Figs. 1 and 2; for a summary of vetulicolians see ref. 6 ). The soft anatomy of these animals is largely unknown, but the anterior part of Vetulicola has been hypothesised to comprise a pharynx with gill-like structures that flexed by means of horizontal and longitudinal muscle fibres attached to four flexible plates covered by a thin outer membrane (see below).
Vermilituus gregarius occurs in four specimens of Vetulicola cuneata from the Chengjiang region (Figs. 2, 4, and 5), plus six specimens of Vetulicola rectangulata from the Haikou region ( Overall, at least 400 specimens of V. rectangulata and 80 specimens of V. cuneata have been collected from the Chengjiang biota (YKLP and CJHMD collections), meaning that Vermilituus gregarius is a rare associate of vetulicolians. Vetulicolian fossils occur as composite moulds where the rock splits through the specimen, each part containing components of both the external and internal surfaces ( Supplementary Fig. 3). For the anterior part of Vetulicola, we interpret Vermilituus gregarius as occupying the space between the interior of the exoskeleton, and the convex surface that appears to demarcate the position of the internal anatomy ( The number of Vermilituus gregarius per associated vetulicolian is variable, ranging from a single tube to 88 individuals (Supplementary data file), and in some cases, V. gregarius occurs in local aggregates of up to 25 individuals, for example in YKLP 13075 (Fig. 3c, e, f). In most cases where V. gregarius aggregates, the individuals are discrete, but occasionally some overlap. The overall size of V. gregarius is from 0.8 to 7.2 mm in length, with maximum diameter ranging from 0.4 to 1 mm (proximal width is circa 0.2 mm). Average tube length varies within individual Vetulicola specimens (Supplementary data file), by a minimum of 1.6 mm (specimens associated with CJHMD 00031) to a maximum of 6.4 mm (specimens associated with YKLP 13073).
Rather than representing post-mortem assemblages, or the result of Vermilituus scavenging or colonising vetulicolian carcasses, all evidence suggests that Vermilituus attached to the body surfaces of living vetulicolians. All infested vetulicolians are preserved within 'event beds' 5 . This means that they were rapidly buried by sediment, and therefore post-mortem colonisation at the seabed is highly unlikely. Tubes of Vermilituus gregarius occur almost exclusively inside the vetulicolians, rather than the external body surface, and preferentially within the anterior part  Fig. 3d). In this rare case, V. gregarius may have over-spilled onto the external surface of the animal or has been displaced post-mortem. Among those in the anterior part, most are located in the convex area between the central groove and the fin-like margins, with some concentrations often in the anterodorsal region ( Fig. 3a, b). Only a few tubes of V. gregarius occur along the margins of Vetulicola. In one case, at least 10 U-shaped tubes grow with a posterior orientation in Vetulicola YKLP 10906 ( Supplementary Fig. 1e). In Vetulicola YKLP 13075, there is a clear association of 24 V. gregarius with the central groove ( Fig. 3f), each having a distinctive orientation with the narrow end of the tube pointing towards the groove.
The consistent occurrence of Vermilituus gregarius inside the anterior section of vetulicolians, combined with the observed patterns of localisation and occasional preferred orientation (Supplementary data file), argues against a chance post-mortem association, or generalist epibiontic habit. In the latter scenarios, the posterior section should also be infested. Furthermore, V. gregarius is absent from any other fossil organism in the Chengjiang biota, suggesting a highly specific relationship. The robust (possibly biomineralised) and curved tubes of V. gregarius . c Paratype YKLP 13084, partial 3D with well-preserved annulation, J-shape morphology. d, h YKLP 13086 under direct light (d) and fluorescence light (h), white arrow shows possible soft tissues. e Paratypes YKLP 13082 and 13083, preserved in 3D with annulation visible proximally: sinusoidal shape and J-shape morphology, respectively (the latter is broken distally and shows sediment fill). f Paratype YKLP 13085, partial 3D with well-preserved annulation, sinusoidal morphology. g, i-k Scanning electron microscopy images. g YKLP 13087 with J-shape morphology. i, j YKLP 13088, boxed area in "i" shows possible paired soft tissues at the termination, magnified in "j". k YKLP 13089, with possible paired soft tissues at the terminal end. Scale bars: a-d, g-i, k, 500 μm; e, f, 1 mm; j, 200 μm.
are consistent with a sessile, attached ecology, but not with a motile scavenger that might have fed on vetulicolians after death. The size range of Vermilituus on each specimen (Supplementary data file) suggests animals growing in situ for some time, rather than colonising carrion. In addition, the lack of evidence for decay and disarticulation of infested vetulicolians combined with their preservation in event beds supports an in vivo association. In this light, the observed patterns in size, number and distribution of V. gregarius tubes also shed light on vetulicolian biology and the ecological relationship between the taxa.

Discussion
Based on their streamlined anterior section and tail-like posterior section (Fig. 6), vetulicolians have been interpreted as actively swimming animals 5 , and some species may have been filter feeders 5,7-9 , using the anterior section-which has been interpreted as a flexible pharynx-to gather food 9 . Vetulicolians have also been interpreted as nektobenthic consumers, feeding selectively at the sediment surface 5 . The host-symbiont relationships for Vetulicola rectangulata and V. cuneata described here support the idea of filter feeding by movement of water through the anterior section of Vetulicola 9 , but the attachment of robust tubes of V. gregarius is not consistent with the hypothesis of active pumping of water facilitated via a flexible body wall that was flexed by muscles. Our interpretation would support the presence of a space between the internal soft tissues and the inner surface of the exoskeleton of Vetulicola to accommodate endosymbionts, and also a degree of rigidity to the exoskeleton.
The earliest marine ecologies with benthic organisms constructing tubular structures are those of the terminal Neoproterozoic, and these likely included suspension-feeding animals 10,11 . None of these organisms constructed tubes that are similar to V. gregarius (Supplementary text), and none had a close symbiotic relationship encrusting a mobile host. Early Cambrian ecosystems, in contrast to those of the terminal Neoproterozoic, were populated by mobile, skeleton-bearing animals. These Cambrian ecosystems are known to preserve symbiotic associations, but they appear to have been dominated by epibiont suspension-feeding animals, most notably brachiopods, with a general absence of endosymbionts until the Ordovician 3 . Therefore, the endosymbiotic relationship between vetulicolians and V. gregarius adds a new dimension to Cambrian ecologies, with an unsuspected intensification of interactions early in the Phanerozoic.
Vermilituus gregarius tubes do not appear to represent the actions of animals scavenging on the Vetulicola carcass postmortem, as encrusting tube-bearing animals are sedentary. Postmortem colonisation of Vetulicola may be considered, as the interior of the anterior section may have provided a protective habitat from predators and a stable surface for encrustation. However, if V. gregarius represents a filter-feeding animal, this would only facilitate food collection if the vetulicolian carcass was drifting through the water column for a protracted period after death. In addition, the vetulicolian hosts show no evidence of decay and/or disarticulation, which would be expected if V.
gregarius were simply exploiting dead and rotting carcasses for food, shelter or a substrate to settle on, whether in the water column or on the seafloor. The seafloor scenario is also unlikely because those vetulicolians that are infested were rapidly buried in 'event beds' 5 .
Accepting a motile nektonic mode of life for vetulicolans, it follows that Vermilituus gregarius most likely infested the surfaces gregarius, including three aggregates of between 10 and 25 specimens (e.g., seen in "e"), and concentration of 24 specimens along the central groove (close-up in "f"): note that these are oriented with the narrow end associated with the groove. This specimen also shows four specimens in the tail ("d"). g, h YKLP 13074, right view of the internal mould of the anterior section infested with about 52 V. gregarius that form aggregates of between 5 and 14, including those that preserve annulation ("h"). Scale bars: a, c, g, 1 mm; b, d, e, f, h 2 mm. COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-01244-1 ARTICLE COMMUNICATIONS BIOLOGY | (2020) 3:517 | https://doi.org/10.1038/s42003-020-01244-1 | www.nature.com/commsbio of Vetulicola from the water column, presumably as gregarious larvae that settled mainly within the anterior section of the host rather than on its surface. This is suggested by the similar fidelity of preservation of both tubes and hosts, suggesting co-habitation (Supplementary data file). Although individual aggregations of V. gregarius often contain specimens of a similar size, the overall size range in each vetulicolian suggests that larvae landed over a protracted period (Supplementary data file), and thus grew over some time. Aggregations do not appear to be the product of asexual budding, as in most cases the individual tubes are discrete (Figs. 1-5, Supplementary Figs. 1 and 2).
Vermilituus gregarius is of uncertain taxonomic affinity (Supplementary text), but the paired, crescent-shaped structures seen protruding at the terminal end of some specimens (Fig. 1h-k) may be remnants of a feeding apparatus that might have been tentacular arms or a lophophore. The overall morphology of V. gregarius is convergent with many living and fossil tubular annelids and possible lophophorates with a filter-feeding mode of life 12 . The aggregation of many V. gregarius within the anterior section of the vetulicolians (Figs. 2-5, Supplementary Figs. 1 and 2) may suggest organisms utilising water currents (cf. oriented cornulitids on Ordovician brachiopods 13 ) generated as part of the host's feeding mechanism or by the host's swimming direction (Supplementary Fig. 4). Patterns of aggregation and orientation occur locally, for example, the concentration of V. gregarius along the central groove in YKLP 13075 (Fig. 3f) and CJHMD 00031 (Fig. 2c, f), which may have been associated with water expulsion 9 . But the absence of a consistent pattern of orientation for V. gregarius may suggest complexity to the internal soft anatomy of vetulicolians, complex patterns of water circulation or that V. gregarius adopted a different mechanism of acquiring food. It is notable that V. gregarius has not been recorded from any other exoskeleton-bearing motile Chengjiang animal, including the many bivalved arthropods that present potential external and internal spaces for colonisation. Again, this suggests a high degree of specificity between host and symbiont.
Aggregative behaviour is a symbiotic strategy used by many tube-bearing animals 14 . The symbiotic association of V. gregarius with vetulicolians is reminiscent of some fossil associations of serpulids that are interpreted to be commensal or mutualistic, for example, in corals and foraminifera 15,16 . The relationship between vetulicolians and V. gregarius may have been commensal, the host providing a safe domicile, and the means of transport and distribution. Nevertheless, that four Vetulicola each carry over 45 V. gregarius suggests that infestation might have been deleterious to the survival of the host in some cases (Fig. 3a, b, g,  h), and that V. gregarius could therefore be parasitic. Competition for resources is particularly problematic if the host and parasite have similar preferences and clearance rates of consumed particles 17 . It is also possible that V. gregarius would compete for oxygen, if the majority were residing alongside the host's respiratory apparatus. In addition, the tube-dwelling life mode of V. gregarius suggests a more complex relationship than simple parasitism, as endoparasites do not typically require protective structures beyond those provided by the host.
The presence of even a small number of V. gregarius tubes might have had a significant fouling effect on surfaces exposed to water flow (see ref. 9 ). Parasitic fouling of otherwise regular structures would increase their surface roughness and therefore the hydrodynamic drag of fluid moving through the body cavity 18 . Growing populations of V. gregarius within the host organism might not only impede fluid through-flow, but the additional weight of >45 individuals could have reduced the swimming efficiency of the host organism and its ability to feed (e.g., see Fig. 3a, b, g, h).
Infestation of the interior area of the anterior section in V. rectangulata may have been encouraged by the wide anterior gape in its exoskeleton. Vetulicola cuneata also possesses extended liplike structures at its anterior opening 19 , allowing larval V. gregarius to enter, whilst other vetulicolians such as Didazoon haoae and Pomatrum ventralis-which have no reported infestationhave narrow anterior openings that may have helped prevent infestation [20][21][22] .
Together with the priapulid worms Mafangscolex sinensis and Cricocosmia jinningensis, which are infested by aggregates of the small worm-like Inquicus fellatus 23 , Vetulicola rectangulata and V. cuneata represent two of four species with host-specific infestation documented in the Chengjiang biota. Of eight other vetulicolid species in the biota 6 , none are infested. Infestation rates in Vetulicola (10 individuals infested from circa 480 specimens overall) are low, representing a rate of about 2%, but this is within the range of, for example, modern polychaete infestations 24 . All infested V. rectangulata are from three localities in the Haikou region, whilst the infested specimens of V. cuneata are from Chengjiang County (Supplementary data file). Thus, V. gregarius had a wide range across the marine basin of the Chengjiang biota 5 . The presence of V. gregarius within the exoskeletons of two host Vetulicola species suggests a propensity for host shift that is also demonstrated in worm infestations of Chengjiang priapulids 24 .
Limited and conflicting evidence of the anatomy of vetulicolians means that their wider taxonomic affinities with other bilaterians are poorly understood. Thus, vetulicolians have been proposed to have affinities with arthropods 19,25 , kinorhynchs 8 , stem-group deuterostomes 9,22 and chordates 26,27 . Whatever their relationships to other bilaterian groups 28 , the presence of a second group of animals in the Chengjiang biota with host-specific infestation and host shift (see also ref. 23 ) indicates that such ecologies were likely widespread in Cambrian ecosystems. Early Cambrian ecosystems preserve rare symbiotic associations 4,29 , but the fossil record shows a bias towards epibiont suspensionfeeding animals, most notably brachiopods, with a general absence of endosymbionts until the Ordovician 3 : this is suggested to be due to reduced predation pressure in Cambrian ecosystems. Contrary to this, our data show that endosymbiotic and fouling relationships between species formed a component of the diversification of animal-rich ecosystems at the beginning of the  Phanerozoic Eon. Given that fouling strategies may substantially change the viability of the host, for example, through physical damage, reduced food supply and mechanical interference (ref. 17 and references therein), fouling may have been a substantial, and hitherto unrecognised driver of evolution in Earth's earliest animal-rich marine ecosystems.

Methods
Materials. The fossils were collected from outcrops of the Yu'anshan Member, Chiungchussu Formation, Eoredlichia-Wutingaspis trilobite Biozone, Cambrian Series 2, Stage 3, Yunnan Province, China (Supplementary data file). All specimens are from 'event beds' 5 . Some specimens were prepared mechanically with needles under a stereomicroscope.
Photography. The digital images of the specimens were captured with a Canon EOS 5D SR camera mounted with Caron MP-E 65 mm (1-5×) or Canon EF 100mm macro lenses under cross-polarised light, and were processed in Adobe Photoshop CC 2018. All measurements were processed with ImageJ version 1.49.
Scanning electron microscope and energy-dispersive X-ray spectroscopy. Scanning electron microscope (SEM) images (Fig. 1g, i-k) were obtained using a Hitachi S3600N at 10-kV accelerating voltage and 30-Pa chamber pressure. Energy-dispersive X-ray spectroscopy (EDX) was performed using an Oxford INCA 350 EDX at 15-kV accelerating voltage and 30-Pa chamber pressure. No evidence for phosphorus, calcium or carbon was identified in our analysis of the tubular shells, which are largely preserved by iron oxide after pyrite, a style of preservation that is widespread in Chengjiang fossils 30 .
Nomenclatural acts. This published work and the nomenclatural act it contains has been registered in ZooBank, the proposed online registration system for the International Code of Zoological Nomenclature (ICZN). The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank. org/". The LSID for this publication is urn:lsid:zoobank.org:pub:4C4CD10E-3C65-4900-8CE9-1A1DC35E2B33.