Exceptionally preserved Cambrian loriciferans and the early animal invasion of the meiobenthos

Microscopic animals that live among and between sediment grains (meiobenthic metazoans) are key constituents of modern aquatic ecosystems, but are effectively absent from the fossil record. We describe an assemblage of microscopic fossil loriciferans (Ecdysozoa, Loricifera) from the late Cambrian Deadwood Formation of western Canada. The fossils share a characteristic head structure and minute adult body size (~300 μm) with modern loriciferans, indicating the early evolution and subsequent conservation of an obligate, permanently meiobenthic lifestyle. The unsuspected fossilization potential of such small animals in marine mudstones offers a new search image for the earliest ecdysozoans and other animals, although the anatomical complexity of loriciferans points to their evolutionary miniaturization from a larger-bodied ancestor. The invasion of animals into ecospace that was previously monopolized by protists will have contributed considerably to the revolutionary geobiological feedbacks of the Proterozoic/Phanerozoic transition. It is unclear when multicellular animals first invaded the microscopic ecological niche between sediment grains given the absence of such animals from the fossil record. Microscopic Loriciferans are described from the Cambrian period, showing an early occupation of this important niche.


Horizon and locality. Five sample horizons in the Deadwood
Formation from across four geographically widespread drill cores in southern Saskatchewan, Canada; dated biostratigraphically to the late Cambrian Furongian epoch (~485-497 million years ago (Ma)). All specimens occur as small carbonaceous fossils (SCFs) recovered from unoxidized mudstone horizons using a low-manipulation HF extraction process 7 . Fossilization occurred in a shallow marine, muddy/ sandy shelf environment (shoreface to offshore) located in a notably more oxygenated, epicratonic setting than classical Burgess Shale-type environments (see Supplementary Fig. 1 for further details). Diagnosis. Body with lorica (posterior 'corset-like' region) and introvert (eversible spiny 'head'); lorica length 120-215 μ m; total length including introvert ~180-300 μ m. Introvert almost as wide as the anterior lorica and around half the length of the lorica when everted. Introvert with > 200 elongate scalids, including robust, jointed and reflexed anterior spinoscalids, and more gracile spinoscalids and/or trichoscalids posteriorly, some with fringed tips. Mouth region delimited by approximately six short clavoscalids or mouth cone elements. Lorica constructed from 20 longitudinal strip-like plicae, each 5-10 μ m wide with squared or bluntly rounded anterior margins ornamented with a fine external fringe of 1.5 μ m long fimbriae; plicae undivided and undifferentiated, becoming poorly defined posteriorly. Lorica with a circular aperture and a slight curvature towards the (presumed) ventral side.

Discussion
The large number of scalids (> 200) on the introvert of E. deadwoodensis, together with the highly elongate scalid morphologies ( Fig. 1d-g) and loricate body, are uniquely shared with modern loriciferans, a recently discovered phylum of microscopic marine metazoans 8 . Just as in modern species 9 , the introvert scalids of E. deadwoodensis exhibit anterior-posterior differentiation with notably robust, jointed and reflexed frontal scalids (spinoscalids) (Fig. 1f), followed by more gracile posterior spinoscalids and/or trichoscalids (Fig. 1g) and Rugiloricus (which has a higher plica count of 30-60, but shares with the fossils a lack of differentiation among the plicae) 9 . All modern pliciloricids have an approximately circular lorica cross-section, in common with the fossil specimens (which exhibit an oval aperture regardless of the angle of flattening; Fig. 2d-f). In contrast, modern loriciferans belonging to Nanaloricidae are characterized by relatively flattened loricae composed of 6-10 strongly differentiated plates 9 . Together with the loriciferan synapomorphies exhibited by the introvert, the preserved lorica characters are consistent with a phylogenetic position for E. deadwoodensis among pliciloricid loriciferans (Supplementary Table 1) and therefore with a crowngroup loriciferan position overall. Alternatively, if pliciloricid-type loricae are plesiomorphic for loriciferans, E. deadwoodensis could conceivably lie in the loriciferan stem (see Supplementary Fig. 3). A stem-group position could explain the absence of differentiated larval stages (such as a Higgins-larva, with its specialized lorica structures), if not simply a result of differential decay or a speciesspecific trait. Another possible plesiomorphic character is the unusual perioral armature in E. deadwoodensis: its ring of approximately six robust structures (possibly clavoscalids or mouth cone elements) compare in number (but not shape) to the six oral styles of adult nanaloricids and the six oral teeth of larval pliciloricids 9 . Ultimately, these characters will help to place E. deadwoodensis once the interrelationships of modern loriciferans are better resolved.
Besides the striking anatomical correspondence to modern loriciferans, E. deadwoodensis is remarkable for its microscopically small body size. On the basis of the scaling relationship between the most intact specimen (lorica length 135 μ m; overall length including introvert < 200 μ m) and the longest lorica (~215 μ m), we estimate a maximum overall body length of less than 300 μ m. A taphonomically imposed upper limit to body size in the fossil populations can be ruled out by the fact that larger (up to millimetre size) small carbonaceous fossils co-occur in each of the five loriciferan-bearing samples. The presence of elongate scalid morphologies in the fossils, along with the absence of posterior lorica projections (tubuli, setae or locomotory appendages), indicates that the fossils are adults, despite their small size. Their size is typical of modern adult loriciferans, which range between 115 and 425 μ m in maximum dimension 9 ; even the largest known 'giant' larvae of Titaniloricus are only 800 μ m long 10 (see Table 1).

Phylogenetic significance.
Establishing the evolutionary significance of minute body size in loriciferans requires consideration of their wider phylogenetic relationships. A suite of morphological characters supports a grouping of loriciferans, kinorhynchs and priapulids in a clade (Scalidophora) within the Ecdysozoa, to the exclusion of panarthropods and nematoids 5,[11][12][13] . A lorica is present during the larval (but never the adult) stages of all but one extant priapulid species, suggesting a close relationship between the two phyla. However, some characters support a closer relationship with kinorhynchs or even nematomorphs 8 , which would imply additional

NaTurE ECoLogy & EvoLuTioN
losses or the convergent evolution of a lorica. Loriciferans have seldom been included in molecular phylogenetic analyses and preliminary data sets that incorporate loriciferan 18S + /− 28S rRNA sequences [14][15][16] have thus far failed to recover a monophyletic Scalidophora ( Supplementary Fig. 3), although the results are probably distorted by long branch attraction 12 . The picture is further complicated by a range of problematic Cambrian fossils that possess a lorica-like structure but are quite unlike modern loriciferans. Sirilorica from the early Cambrian (Series 2) Sirius Passet biota of North Greenland, for example, has a bizarrely interdigitated 'lorica' some 70 mm long, orders of magnitude larger than those found in any living loriciferan 17 .
More conventional lorica-like structures are present in comparably macroscopic (up to 25 mm) Palaeopriapulites, Protopriapulites and Sicyophorus from the early Cambrian (Series 2) Chengjiang biota of south China 18,19 (Table 1), although their priapulid-type introverts with comparatively few, short scalids 13 gives them the appearance of giant priapulid larvae; indeed, they commonly group with priapulids in cladistic analyses 11 . Middle Cambrian (Series 3) Orstenoloricus, known from incomplete specimens lacking an introvert, has also been compared to loriciferans 20 , although the concertina-like folding of its lorica, extensive neck region and millimetric body size (Table 1) ally it more closely with larvae of the extant priapulid Tubiluchus.
In contrast, E. deadwoodensis provides the first unambiguous evidence for microscopic loriciferans in the fossil record and establishes their long evolutionary history independent of priapulids and other ecdysozoans. At the same time, it exhibits lorica characters that can be interpreted as plesiomorphic for both priapulids and loriciferans: a lorica with a circular cross-section also occurs in the earliest loricate larval stages of Priapulus 13 , in the candidate 'basal' priapulid Tubiluchus 13,21 and in the 'giant' Cambrian loricate fossils [17][18][19] . To this extent, E. deadwoodensis supports the homology of loricae between the two phyla. It has been suggested that loriciferans derive from a largerbodied, priapulid-like ancestor via progenesis from a form with a loricate larva (that is, precocious sexual maturity leading to retention of a larval morphology) 22 . The presence in the Cambrian fossil record of macroscopic loricate forms offers an alternative mechanism, that of secondary miniaturization 3,23 from a large loricate adult. Either way, a larger-bodied ancestor for microscopic loriciferans is predicted by their anatomical complexity, most strikingly expressed by their intricate myoanatomy 24 , hundreds of innervated scalids and thousands of cells 1 . Indeed, a parsimonious reconstruction of lorica evolution suggests that E. deadwoodensis and modern loriciferans are derived from a larger-bodied (millimetric, centimetric or even decimetric) ancestor, given the stem-loriciferan, stem-priapulid and/or deeper positions for other Cambrian loricate worms and the macroscopic size of nonloricate outgroups (see Supplementary Fig. 3).
A hypothesis that minute animals existed for tens or hundreds of millions of years before the Cambrian explosion 6,25 is attractive, in that it reconciles deep molecular clock predications of divergences 4,5 with the well-established empirical pattern from the fossil record. A long pre-Cambrian history of microscopic loriciferans seems unlikely, however, considering their derived phylogenetic position and a lack of plausible (that is, macroscopic) ancestors before the Cambrian (or identifiable animal traces before the latest Ediacaran) 26 . At a deeper level, ambiguities in the metazoan tree allow for either a large-bodied 27 or small-bodied 2 ecdysozoan ancestor. The discovery of E. deadwoodensis demonstrates the potential for even the tiniest cuticularized metazoans to be preserved as small carbonaceous fossils, augmenting the existing SCF record of comparatively tough cuticular sclerites of larger-bodied relatives 7,28 . The absence of any such cuticular remains before the Cambrian points to a pronounced late-stage radiation of ecdysozoans, challenging molecular clock predictions 4,5 of > 630 Ma or > 580 Ma intra-ecdysozoan divergences.
Early evolution of the metazoan meiobenthos. The exceptionally small body size of modern loriciferans, which are among the smallest known metazoans, is attributed to their specialized ecologies as members of the obligate, permanent metazoan meiobenthos; that is, those animals that spend their entire life cycle living interstitially between sand grains or among finer-grained sediments on the seafloor 9 . Following their first formal description 8 in 1983, loriciferans have been discovered in a wide range of habitats, from the intertidal zone to > 8 km-deep ocean trenches and from sandy sediments to anoxic muds 9 . Even so, they consistently exhibit key ecological traits that are linked to microscopic body size 29 , including the absence of a planktonic larval stage, asymptotic growth, (weak) motility and a particle-discriminating (apparently bacterivorous or microalgal) diet 9 . A comparable ecology for E. deadwoodensis is supported by the equivalence of scale, the restricted size range in the fossil populations and the various scalid adaptations linked to feeding and locomotion.
Modern meiobenthic organisms are distinguished operationally 1 from larger macrobenthos and smaller microbenthos if they pass through a mesh of 1 mm (or sometimes 500 μ m), but not one of 63 μ m (or 45 μ m). By this measure, the larvae of various Cambrian animals would have been temporary members of the meiobenthos. Identifying fossils as members of the permanent meiobenthos, however, requires additional evidence of both the adult anatomy and benthic ecology. With its adult-type lorica, introvert armature and necessarily bottomdwelling, non-swimming ecology, E. deadwoodensis readily passes these tests, but the case is less clear for other Cambrian candidates. Fossils from phosphatized 'Orsten-type' assemblages often exhibit meiobenthic dimensions, including a variety of panarthropods 30,31 , scalidophorans 32,33 and cnidarians 34 , but most of these forms are

NaTurE ECoLogy & EvoLuTioN
known to have grown to at least millimetric adult body sizes (Table 1). One possible exception is a single 350 μ m-long larval tardigrade 30 that may have retained its small size as an adult. A few of the Orsten pancrustaceans might also fall into this category, in particular Bredocaris, which has a maximum known (?adult) body length of 850 μ m and Skara, which is somewhat longer (1.2 mm), but has an elongate body similar to some modern meiobenthic crustaceans 31 . In any event, E. deadwoodensis is the smallest animal yet identified in the Cambrian fossil record and the clearest evidence for early animals living permanently among and between sedimentary grains. The discovery of E. deadwoodensis establishes the presence of a specialized metazoan component in the meiobenthos by at least the late Cambrian, adding a fundamental new dimension to the geobiological feedbacks associated with the Cambrian explosion. In the modern world, meiobenthic animals can be enormously abundant in marine sediments (10 6 -10 7 individuals per square metre), constituting a substantial fraction of sedimentary biomass and profoundly influencing local sedimentological and biogeochemical expression, including sediment cohesion, solute transport and organic matter cycling 1,[35][36][37][38][39] . In turn, the 'engineering' effects of meiobenthic animals inevitably shape the surrounding microbenthic (smaller protistan) and macrobenthic communities, just as the activities of macroscopic animals locally either promote or suppress meiobenthic abundance and diversity 1,36,38 .
Not all small interstitial organisms are animals, of course, and there is a considerable degree of overlap in size and ecosystem function with larger benthic protists, most notably ciliates and foraminifera 1,25 . Moreover, protistan-grade forms were probably established well before the rise of animals, as suggested by accumulations of testate amoebae in mid-Neoproterozoic (~750 Ma) strata 40 . Importantly, however, the habits and geobiological impacts of these two components of the benthic meiofauna can be profoundly different. Organgrade metazoans express a fundamentally greater repertoire of habits and interactions than their unicellular counterparts, not least the dramatically enhanced levels of motility/dexterity (conferred by a bilaterian musculoskeletal system) and the revolutionary impact of processing food via a through-gut. The Cambrian colonization of meiobenthic habitats by metazoans clearly attests to this adaptive potential. At one level it presumably entailed a competitive displacement of incumbent protists, although the accompanying engineering effects may also have facilitated subsequent protistan radiations.
Among the most obvious and characteristic features of the Cambrian explosion is a dramatic increase in maximum body size, culminating in metre-long anomalocaridids 41 . No less impressive, considering the competitive demands on early meiobenthic pioneers, is the extremely small size of Cambrian loriciferans. The size increase through this interval is often viewed as an exploration of vacant ecospace coincident with a permissive rise in ambient oxygen levels 4 , but extremes in body size can also indicate strong selective displacement away from more optimal ranges 42 . Miniaturization in a Cambrian loriciferan to near the lower limit for free-living animals implies that the prevailing benthic ecology was sufficiently escalated to drive novel defensive strategies (via a small size 'refuge' or lifehistory modification) and/or the exploitation of a rich but otherwise inaccessible food source 23 . Certainly, the advent of burrowing by larger metazoans in the Cambrian resulted in a systematic restructuring of the benthic environment by converting matgrounds to mixgrounds 43 , potentially opening new ecospace for smaller metazoans via increased sediment oxygenation. A wider analysis, however, suggests that there is no straightforward relationship between oxygen, macrofaunal burrowing and meiobenthic opportunities: seemingly inhospitable modern-day microbial mat environments can host abundant and diverse interstitial animals as well as protists 44 and some extant loriciferans apparently occupy permanently anoxic habitats 45 , albeit as a derived ecology.
E. deadwoodensis documents the evolution of extreme miniaturization in one lineage of Cambrian animals, but there is a conspicuous absence of others, notably nematodes, harpacticoid copepods, turbellarians and micro-annelids, which are by far the most geobiologically important modern groups 1 . Neither the taxo nomic composition nor ecological contribution of the Cambrian meiobenthos was likely to have been fully modern, but the challenge is to distinguish genuine absence from taphonomic loss. Meiobenthic animals have long been of interest to palaeobiologists precisely because of their perceived invisibility as fossils, but not all are impossible to preserve. While molecular phylogenetic analyses 2,3 , preserved benthic biomarkers 46 and sedimentary (bio)fabrics 39,47 provide important constraints on meiobenthic evolution, small carbonaceous fossils offer a direct window onto cryptic animal ecosystems through deep time.
Data availability. The specimens on which this study is based are accessioned in the collections of the Geological Survey of Canada