The Weng'an Biota (Doushantuo Formation): an Ediacaran window on soft-bodied and multicellular microorganisms

The origin and evolutionary assembly of animal body plans comprises one of the most formative episodes in the history of life. Animals are ecosystem engineers and their appearance fundamentally changed our planet's ecology (Butterfield 2011a). Despite the importance of this evolutionary episode, many aspects of the timing and nature of the event remain poorly constrained. Molecular-clock analyses estimate that animals originated by the Cryogenian and diversified through the Ediacaran (Peterson & Butterfield 2005; Erwin et al. 2011; dos Reis et al. 2015), but fossil evidence of animals from before the Cambrian is controversial (Erwin et al. 2011; dos Reis et al. 2015; Cunningham et al. 2017). The Weng'an Biota is one of the few Lagerstätten from the critical interval in which early animals are expected according to molecular-clock studies. In this Ediacaran fossil assemblage, organisms are phosphatized in cellular and even subcellular detail, providing a rare glimpse of soft-bodied and multicellular life at this time. Early research appeared to fulfil expectations of the presence of metazoans with reports of embryonic (Xiao et al. 1998), larval (Chen et al. 2000, 2002) and adult (Xiao et al. 2000; Chen et al. 2002, 2004; Yin et al. 2015) animals from the Weng'an deposit. However, subsequent analyses have cast doubt on this view, and there is currently much disagreement over these interpretations (Bailey et al. 2007a; Huldtgren et al. 2011; Bengtson et al. 2012; L. Chen et al. 2014, Xiao et al. 2014a). Here, we review the stratigraphic position, geological age and environmental setting of the deposit, and present an overview of the biota and an assessment of the phylogenetic affinities of the various taxa.

The Weng'an Biota occurs within the Ediacaran Doushantuo Formation (551 – 635 Ma, Condon et al. 2005) of south China (Fig. 1). In addition to the phosphatized microfossils from Weng'an, this formation has yielded silicified microfossils (Yin et al. 2004) and macrofossils that are preserved as 2D carbonaceous compressions including macro-algae (Xiao et al. 2002) and the putative ctenophore Eoandromeda (Tang et al. 2008, 2011). The Weng'an Biota itself is known from localities in Weng'an County, Guizhou Province. The Doushantuo Formation overlies the Marinoan glacial tillites of the Cryogenian Nantuo Formation that can be dated to 635 Ma (Condon et al. 2005). It is overlain by the Ediacaran Dengying Formation, which contains fossils of the classical Ediacara macrobiota (Sun 1986; Xiao et al. 2005; Z. Chen et al. 2014). The base of the Dengying Formation can be dated to 551 Ma (Condon et al. 2005). In Weng'an, the Doushantuo Formation is composed of five units that have been described in detail by Xiao et al. (2014b) and Yin et al. (2015). The Weng'an Biota occurs mainly in Unit 4, the Upper Phosphorite Member, but also in Unit 5. Unit 4 is divided into 4A, a black phosphorite, and 4B, a grey dolomitic phosphorite (Dornbos et al. 2006; Xiao et al. 2014a,b; Yin et al. 2015).

The age of the biota has been debated (Budd 2008; Erwin & Valentine 2013; Xiao et al. 2014a; Yin et al. 2015), with arguments focusing on the correlation of two karstic surfaces, one at the top of Unit 3 and the other within Unit 5 (for detailed discussion of Doushantuo correlation see Zhu et al. (2007), Zhu et al. (2013) and Yang et al. (2015)). If the lower surface is correlated to the c. 582 Ma Gaskiers glaciation (Condon et al. 2005) then the biota would be younger than 582 Ma. However, the lower surface may be older (Yin et al. 2015) and, if the upper karstic surface correlates to the Gaskiers glaciation (Xiao et al. 2014a), then the biota would be older than 582 Ma. Direct radiometric dates for Unit 4 at Weng'an have been inconclusive, giving Pb–Pb isochron ages of 572 ± 36 Ma for Unit 4A (Y. Chen et al. 2009) and 599 ± 4 Ma for Unit 4B (Barfod et al. 2002). However, a recent U–Pb date of 609 ± 5 Ma from a tuff immediately above Unit 4 at Zhancunping, in Hubei Province (Zhou et al. 2017), suggests that the Weng'an biota is probably older than 609 ± 5 Ma and probably older than the Gaskiers glaciation, which cannot be related to the karst surface at the top of Unit 3 if this date is correct. Acritarchs identical to those containing embryo-like fossils are found just above an ash band dated to 632.5 ± 0.5 Ma in Doushantuo sections from the Yangtze Gorges (Yin et al. 2007), suggesting that these organisms could have existed at this time. In summary, the biota is probably older than the classical but enigmatic Ediacaran biota and considerably predates the rich animal fossil record of the Cambrian. Putative animals from the assemblage are therefore candidates for the oldest animals in the fossil record.

The Weng'an Biota is interpreted as having been deposited in an outer-shelf environment on a SE-facing passive margin (Jiang et al. 2011). Abundant wave ripples and cross-bedding features indicate deposition above fair-weather wave base (Xiao & Knoll 1999). The fossils were probably deposited in oxic conditions (Shields et al. 2004), although phosphatization may have occurred in anoxic sediments (Muscente et al. 2014; Schiffbauer et al. 2014). The soft-bodied organisms of the biota are three-dimensionally preserved in calcium phosphate and can be preserved at a subcellular level (Hagadorn et al. 2006; Huldtgren et al. 2011). However, even the best-preserved specimens are a complex amalgam of cements, making it challenging to determine which aspects represent preserved biology (Xiao et al. 2000; Bengtson 2003; Bengtson & Budd 2004; Cunningham et al. 2012a; Schiffbauer et al. 2012). The preservation of Weng'an fossils is discussed in Box 1. The fossils occur either as, or within, phosphatic grains that have been abraded and rounded, indicating transport from other parts of the basin after initial preservation (Xiao et al. 2007b). In unit 4A, a c. 5 m thick black phosphorite, the fossils occur in reworked phosphatic clasts. As a result, they cannot be released by acetic acid maceration and have generally been studied in petrographic thin sections (e.g. Chen et al. 2004; L. Chen et al. 2014; see Box 2 for a comparison of 2D and 3D analytical techniques). In unit 4B, a c. 10 m thick grey dolomitic phosphorite, the microfossils are abundant and in places are so concentrated that the layers resemble oolites. This is a grainstone composed largely of microfossils that have been phosphatized before being reworked, transported and winnowed (Xiao & Knoll 1999; Xiao et al. 2007b). The fossils can be extracted by dissolution of the carbonate constituents of rock samples in weak acetic acid and manual sorting of the resulting residues. Specimens can then be studied using scanning electron microscopy (e.g. Xiao et al. 1998) or tomographic techniques (e.g. Hagadorn et al. 2006) and can be re-embedded in resin and sectioned for further petrographic and geochemical analyses (reviewed by Cunningham et al. 2014).

A variety of algal taxa have been reported from Weng'an (Zhao 1986; Zhang 1989; Zhang & Yuan 1992; Xiao et al. 1998, 2004; Xiao 2004). Some simpler forms, such as Archaeophycus (Fig. 2j), show interesting similarities to extant bangialean red algae (Xiao et al. 1998), although other affinities cannot be ruled out (Xiao et al. 2014a). More complex forms such as Thallophyca and Paramecia share characters with floridophyte red algae including pseudoparenchymous construction, differentiated thalli and possible reproductive structures (Xiao et al. 2004). However, the putative algae have received relatively little attention because of the focus on the search for animals. Many specimens lack the overall form of a photosynthetic organism. For example, the blades typical of various seaweeds have not been recovered and it is hard to envisage how cells positioned centrally within globular masses would have functioned in a photosynthetic organism (Fig. 3a and b). Algae may have been used as a wastebasket taxon for assorted irregular forms and rejected animal candidates. These fossils merit further study and there is a prospect that they may include developmental stages of other organisms including the embryo-like fossils, especially given their typically larger size.

The acritarchs (Fig. 2k) from Weng'an have been reviewed comprehensively by Xiao et al. (2014b) and form part of the Doushantuo–Pertatataka microbiota (Zhou et al. 2001, 2007; Liu et al. 2014). The phylogenetic affinities of acritarchs are unknown and they probably represent a polyphyletic assortment of eukaryotes (Huntley et al. 2006). Some Doushantuo acritarchs contain embryo-like fossils (Yin et al. 2007), leading to the suggestion that other Ediacaran acritarchs might be resting cysts of these organisms (Cohen et al. 2009). The interpretation of the Doushantuo–Pertatataka acritarchs is therefore at least partially linked to that of the embryo-like fossils, although they may well represent a polyphyletic assemblage. The affinities of taxa should be considered on a case-by-case basis.

Weng'an fossils that have been interpreted as the embryos of early animals (Xiao et al. 1998) have been the focus of most attention and debate (Fig. 2a–f). They have been interpreted as metazoans (Xiao et al. 1998) including bilaterians (Chen & Chi 2005; J. Chen et al. 2006, 2009a,b; Yin et al. 2013), as stem-group metazoans (Hagadorn et al. 2006; Schiffbauer et al. 2012; L. Chen et al. 2014) or as members of non-metazoan clades (Bailey et al. 2007a,b; Butterfield 2011b; Huldtgren et al. 2011, 2012; L. Chen et al. 2014; Zhang & Pratt 2014). We consider the various claims below.

Bailey et al. (2007a,b) proposed an interesting hypothesis that the embryo-like fossils might be giant sulphur bacteria similar to the living Thiomargarita. These bacteria can be similar in size and shape to the embryo-like fossils from Weng'an and are capable of undergoing at least a few rounds of palintomic division. However, subsequent analyses have shown that these bacteria cannot account for key morphological aspects of the fossils such as the presence of ornamented envelopes, outer acritarch vesicles, and probable lipid vesicles and nuclei (Donoghue 2007; Xiao et al. 2007b; Huldtgren et al. 2011; Cunningham et al. 2012b). Moreover, evidence from experimental taphonomy showed that Thiomargarita cells are not replicated by biofilm-forming bacteria, meaning that they do not form the stable bacterial pseudomorphs that are thought to be the precursor to exceptional phosphatization (Cunningham et al. 2012b; see Box 1).

Reports of embryonic bilaterians do not withstand scrutiny. Some are based on identifications of cell geometries argued to be unique to bilaterians. These include specimens purported to preserve endodermal cords (J. Chen et al. 2009b), polar lobes (J. Chen et al. 2006, 2009a; Yin et al. 2013), embryonic polarity (J. Chen et al. 2009a,b) and duet cleavage (J. Chen et al. 2009a). In each case, the specimens can be alternatively interpreted as examples of embryo-like fossils that have undergone taphonomic and diagenetic processes (Huldtgren et al. 2011; Cunningham et al. 2012a). Specimens interpreted as possible bilaterian or cnidarian gastrulae and larvae (Chen et al. 2000, 2002; Chen & Chi 2005) are more probably deformed cysts filled with phosphatic cements (Xiao et al. 2000). The presence of meroblastic embryos (Yin et al. 2016) would represent the long sought-after confirmation of the presence of animal embryos. A possible alternative interpretation of these specimens, which are associated with a large population known to undergo asynchronous division (Hagadorn et al. 2006), is that they have undergone unequal division. It is currently difficult to test between these possible interpretations.

The embryo-like fossils probably represent one taxon dividing from a single cell to thousands of cells (Fig. 2a–f). This taxon has been named either Tianzhushania or Megasphaera according to different taxonomic interpretations (Yin et al. 2004; Xiao et al. 2014b). Tianzhushania is preferred here (see Box 3). These specimens were interpreted as animal embryos by Xiao et al. (1998) based on the similar size and the presence of palintomic cell division, Y-shaped junctions between cells and an ornate enclosing envelope. More recently, these have been considered as stem, rather than crown, animals because later stages lack evidence for epithelial organization, which is characteristic of modern embryos (Hagadorn et al. 2006). However, the placement of these fossils in the animal total-group has also been questioned (Bailey et al. 2007a; Butterfield 2011b; Huldtgren et al. 2011; L. Chen et al. 2014; Zhang & Pratt 2014).

None of the characters that have been used to justify an animal interpretation are exclusive to animals (see Fig. 4). Features such as palintomic cleavage, Y-shaped cell junctions and an ornate envelope are found in non-animal groups (Huldtgren et al. 2011, 2012). They are therefore consistent with an animal interpretation, but they are not diagnostic characters. They are insufficient, either in isolation or in combination, to justify an animal affinity.

It has also been suggested that Tianzhushania exhibits characters gained in the animal stem lineage. L. Chen et al. (2014) described discrete clusters of cells (‘matryoshkas’) within embryo-like fossils with hundreds of cells. They interpreted these as reproductive propagules and presented them as evidence for spatial cell differentiation, germ–soma separation and apoptosis. Based on these characters, along with functional cell adhesion, obligate multicellularity and the potential lack of a rigid cell wall, L. Chen et al. (2014) argued that Tianzhushania might be a stem-animal that had gained some, but not all, of the characters that are present in animals, but not choanoflagellates. The occurrence of dividing cells within the embryo-like fossils is merely an expectation of the existing observation that they exhibit asynchronous cell division (Hagadorn et al. 2006). Nevertheless, the interpretation of differentiation and germ–soma separation requires that the clusters are part of the embryo-like organism rather than an exogenous parasite. L. Chen et al. (2014) suggested that there is a developmental continuation from the typical cells of these specimens and the clusters, which rules out an exogenous origin. However, there is a discontinuity between monads, dyads and tetrads, which show evidence for palintomy, and larger ‘matryoshka’ clusters, which do not (Tang 2015; Cunningham et al. 2016). Although there must be a switch from palintomy at some point (Chen et al. 2017), the lack of intermediates weakens the evidence for an endogenous origin (Tang 2015; Cunningham et al. 2016).

The argument for apoptosis is that, because the enclosing envelope imposes a constant volume throughout ontogeny, the only way to create the space required for the putative reproductive structures to grow is for other cells to die off. However, the volume of Tianzhushania may well not be constrained in this way, given that early stages often do not occupy the full envelope, and putative later stages provide evidence for distension and then rupture of the envelope (Liu et al. 2009; Huldtgren et al. 2011). The evidence for differentiation, germ–soma separation and programmed cell death is therefore unconvincing. Moreover, there is also uncertainty regarding the other proposed animal characters. There is no evidence that the walls of Tianzhushania are any less rigid than those of non-animal groups such as non-metazoan holozoans (Marshall & Berbee 2011), amoebozoans (Olive 1975), ciliates (Park et al. 2005) or volvocalean algae (Hallmann 2006), which also have Y-shaped cell junctions. Nor is there evidence that cell adhesion is different from that seen in groups such as non-metazoan holozoans (volvocalean algae achieve adhesion through cytoplasmic bridges, which are absent in the fossils).

To summarize, Tianzhushania does not exhibit characters that are sufficient to identify it as an animal. Evidence for the presence of characters gained in the animal stem lineage is equivocal. There is therefore no justification for an interpretation of Tianzhushania as an animal, although a stem-animal affinity cannot be definitively rejected on the basis of current data (Huldtgren et al. 2011, 2012). Above all, the available evidence does not yet allow these fossils to be used to substantiate hypotheses on the timing of animal diversification.

A number of alternative interpretations have been proposed, yet the affinities of these organisms remain uncertain. Comparisons with non-metazoan holozoans (Huldtgren et al. 2011), algae (Butterfield 2011b; L. Chen et al. 2014; Zhang & Pratt 2014) or ciliates are plausible but as yet unsubstantiated and require further investigation. The primary factor that has hindered interpretation of these fossils is poor understanding of the later stages of the organisms’ ontogeny. A number of candidates have been proposed, but none are widely accepted. Xiao et al. (2007a) suggested that helically coiled specimens (Fig. 2g), now named Helicoforamina, might be later stages, perhaps representing a coiled vermiform animal. However, this taxon is enigmatic and has also been argued to be an embryo of the ctenophore-like fossil Eoandromeda, which has eight spiral arms (Tang et al. 2008, 2011), or a single-celled stage of Spiralicellula (Fig. 2h), a form that resembles Tianzhushania, but differs in having coiled cells (Huldtgren et al. 2011). Peanut-shaped specimens with hundreds of thousands of cells (Fig. 3c and d) have also been interpreted as later developmental stages (Huldtgren et al. 2011). These have single cells and clusters of cells that are isolated from the main mass of cells and have been argued to be reproductive propagules. L. Chen et al. (2014) have also described specimens with putative reproductive propagules. If Tianzhushania does reproduce via propagules, then this indicates a lifecycle incompatible with at least crown animals. However, in both cases the interpretation as propagules has proven contentious (Xiao et al. 2012; Tang 2015) with a key issue being the incomplete knowledge of the lifecycle of Tianzhushania.

Structures from the Weng'an Biota have been interpreted as siliceous sponge spicules (Li et al. 1998). However, these have been shown by subsequent analysis not to be composed of silica and to lack convincing sponge characters (Zhang et al. 1998; Antcliffe et al. 2014; Muscente et al. 2015). More recently, Eocyathispongia (Fig. 3c) has been described as a possible sponge from Weng'an (Yin et al. 2015). This is known from a single specimen that is preserved at a cellular level. Eocyathispongia is considered to be one of the strongest candidates for a Precambrian sponge. However, although it could be a sponge, it has no convincing sponge apomorphies such as pores or spicules, just a generalized sponge gestalt. More detailed characterization of the anatomy of Eocyathispongia is required. For example, high-resolution tomography might reveal evidence for the presence or absence of sponge characters and help to constrain the affinity of this enigmatic organism.

A group of tubular microfossils assigned to the genera Sinocyclocyclicus (Fig. 3e and f), Ramitubus (Fig. 3g and h), Crassitubus and Quadratitubus have been interpreted as cnidarian-like eumetazoans from the Weng'an Biota (Xiao et al. 2000; Chen et al. 2002; Liu et al. 2008). These genera have regularly spaced cross walls and have been compared with tabulate corals. In a coral-like body plan, the spaces between the cross walls represent the former living positions of the polyp and would be expected to be empty or filled with diagenetic cements. However, the fossils preserve biological structures in these spaces, which is incompatible with a cnidarian interpretation (Cunningham et al. 2015). There is no evidence to support a placement of these tubular fossils within eumetazoans or animals.

Vernanimalcula has been described from thin sections as a miniature, adult bilaterian from the Weng'an Biota (Chen et al. 2004; Petryshyn et al. 2013). It is purported to preserve bilaterian characters such as a mouth, gut, anus and paired coelomic cavities. However, all of the putative bilaterian characters can be alternatively interpreted as artefacts resulting from abiological diagenetic apatite cements, which are ubiquitous in the deposit (Bengtson 2003; Bengtson & Budd 2004; Bengtson et al. 2012; Cunningham et al. 2012a). Moreover, 3D analyses show that Weng'an fossils such as acritarchs and fertilization envelopes that clearly lack bilateral symmetry can exhibit Vernanimalcula-like morphology when sectioned in particular orientations (Bengtson et al. 2012). There is no evidential basis for interpreting Vernanimalcula either as a bilaterian or as an animal of any kind.

Research into the Weng'an Biota is currently in a transitional phase. Early research involved many attempts to demonstrate the long-expected presence of animals, including bilaterians, in the Ediacaran. This has been followed by a spell of critical analysis of these claims that has shown that none of these fossils can so far be confidently identified as stem- or crown-group metazoans. The research is now entering a phase where more targeted analysis of the palaeobiology of each taxon is helping to constrain wide-ranging hypotheses of affinity more rigorously. Many fossils are known only from a few specimens and have received limited attention. One such example is the enigmatic fossil Caveasphaera (Fig. 2i), which has been tentatively compared with cnidarian embryos (Xiao & Knoll 2000) but requires further investigation. Such taxa betray a cryptic diversity that has been overlooked because of the focus on Tianzhushania.

The Ediacaran was a critical interval in the history of life (Butterfield 2007) and the Weng'an Biota offers a unique glimpse of microscopic, multicellular and soft-bodied organisms at this time. It can provide important insights into Ediacaran biology and the evolution of multicellular organisms at this time, possibly including animal-type multicellularity. Future work constraining the diversity, affinity and ontogeny of the Weng'an organisms is necessary before this potential is fully realized.

We thank S. Dornbos and an anonymous referee for constructive comments that helped improve the paper. Data statement: There are no data associated with this paper.