Brachiopods: origin and early history

Despite many major advances in recent years, three key challenges remain in bringing clarity to the early history of the phylum: (1) identifying the origin, morphology and life modes of the first brachiopods; (2) understanding the relationships of the major groups to each other and higher sister taxa; and (3) unravelling the roles of the Cambrian and Ordovician radiations that set the agenda for much of subsequent brachiopod evolution. Since some 95% of all brachiopod taxa are extinct, the fossil record is the primary source of data to frame and test models for the evolution of the phylum. The acquisition of new, and the redescription of existing faunas, in precise spatial and temporal frameworks, using new and well‐established analytical and investigative techniques, are as important as ever.

O U R current understanding of the Cambrian origin and early history of the brachiopods is far from complete; nonetheless the Brachiopoda provides a rich source of data for addressing major research questions relevant to their evolution and that of other invertebrate phyla. Although the monophyletic Brachiopoda remains firmly placed amongst the lophotrochozoan protostomes, the detailed internal topology of the brachiopod clade and its putative sister taxa have advanced significantly since the benchmark analysis of Williams et al. (1996). In particular, the revised brachiopod Treatise volumes (part H, volumes [1][2][3][4][5][6][1997][1998][1999][2000][2001][2002][2003][2004][2005][2006][2007]; see also Carlson 2016) still represent a comprehensive and detailed snapshot of research on and across the phylum. This snapshot, however, has been complemented by recent discoveries of both exceptionally-preserved and skeletal stem group taxa within the time-frame of the Cambrian Explosion, and commonly associated with Lagerst€ atten, prompting a more detailed picture of the basal-most taxa on the brachiopod tree (Zhang et al. 2011a(Zhang et al. , b, 2013Zhang & Holmer 2013). There is also a large body of new data which has a bearing on the broad-frame classification of the phylum, established by Williams et al. (1996) with subsequent and significant modifications (e.g. Carlson 2016) that merit discussion.

THE EARLY PALAEOZOIC FOSSIL RECORD OF BRACHIOPODS
The brachiopods or lamp-shells are a distinctive and diverse group of marine, mainly sessile, benthic invertebrates with a long and varied geological history dating back to the early Cambrian (Fig. 1A). They are one of the few groups of marine animals, which have an enviably complete fossil record, from the emergence of the earliest skeletonized representatives in the early Cambrian (Terreneuvian) to a sporadic distribution in modern oceans (Ushatinskaya 2008;Clausen et al. 2015;Skovsted et al. 2015). Over 12 000 fossil species and approximately 350 living species have been reported, belonging to nearly 6000 genera (see e.g. Harper 2005). Significant information on their relatively simple body anatomy can be extracted from their hard parts: a bivalved shell together with skeletonized brachial supports, including impressions of their muscle system, mantle canals, pedicle and other attachment structures, features of the lophophore, and even some features of the nervous system (e.g. pedicle nerve impressions in lingulids; Holmer et al. 2016). Detailed studies of the brachiopod shell ultrastructure continue to reveal important details about the early evolution of shell secretion within the phylum (e.g. Williams & Cusack 2007;Holmer et al. 2008a). Recent discoveries of exceptionally-preserved specimens with soft parts, which are mostly from the lower Cambrian Chengjiang Konservat Lagerst€ atte of Kunming, southern China, provide an important window into soft body anatomy of the extinct brachiopod groups, close to the initial divergence of major lineages within the phylum, as well as confirming the early origin of the U-shaped gut and the lingulide-type pedicle in linguliforms (Zhang & Holmer 2013). It has also been possible to infer the existence of lophophore and setal structures in the earliest nonbivalved, stem-group brachiopods (e.g. Holmer et al. 2008b). Important advances in understanding the early stages of ontogeny in Early Palaeozoic brachiopods have also been made; the preserved protegulum and larval shell in juveniles preserve important information on the possible feeding habit of the larva, timing and characters of metamorphosis, including the number of larval setal sacs, secretion of protegulum at or prior settlement, characters of initial larval attachment and other characters (Freeman & Lundelius 2005;Popov et al. 2007bPopov et al. , 2009Popov et al. , 2012Bassett et al. 2008). Finally, there has been significant progress in studies of brachiopod molecular phylogeny, although they are still in the pioneering stage, and with sometimes contradictory results (Passamaneck & Halanych 2006;Paps et al. 2009;Hausdorf et al. 2010;Sperling et al. 2011;Nesnidal et al. 2013).

RELATIONSHIPS BETWEEN MAJOR BRACHIOPOD GROUPS
Recent brachiopods are distributed across three major groups, which are currently referred to three subphyla: Linguliformea, Craniiformea and Rhynchonelliformea (Williams et al. 1996). These subphyla may have split from each other at an early stage in brachiopod evolution, and probably prior to the acquisition of a mineralized shell in the early Cambrian (Gorjansky & Popov 1985, 1986Ushatinskaya 2008;Holmer et al. 2011). In any event, the living representatives of these three major lineages are clearly separated from each other in major aspects of shell morphology, soft-body anatomy and early ontogeny in the early fossil record of the group (Carlson 1991a;Popov et al. 1993).
The key characteristics of each subphylum are outlined in Table 1. In addition, each possesses a double row of tentacles throughout their ontogeny, which was previously considered to be an apomorphic feature of linguliform brachiopods (Holmer et al. 1995;Williams et al. 1997). Emig (in Zhang et al. 2009, fig. 4) demonstrated that a single row of tentacles is present on the trocholoph of the discinid Pelagodiscus atlanticus, and thus not a characteristic of only the linguliforms, but most probably plesiomorphic for all groups of the brachiopods. Differences between the three subphyla (Table 1) are apparent by the end of the Ordovician; however, recent studies of Early Palaeozoic brachiopods have revealed a more complex and intricate picture.
A key problem in unravelling the early evolution of major brachiopod clades concerns our understanding of the polarity of morphological characters in phylogenetic analyses. According to some recent molecular studies (e.g. Helmkampf et al. 2008;Nesnidal et al. 2013, fig. 1) brachiopods are placed at the base of lophotrochozoan clade, while in others they represent the most advanced lophotrochozoans (e.g. Paps et al. 2009;Nesnidal et al. 2013, fig. 4); thus any attempt at outgroup selection is a fundamental problem and the available choices (e.g. annelids, molluscs, bryozoans and phoronids) will inevitably influence the result, because all of them may be derived in one way or another in relation to the brachiopods (see e.g. Fig. 1B).

BRACHIOPOD PHYLOGENY: AN OVERVIEW
Brachiopods are commonly considered to be a monophyletic group and, for most of the history of their study, a two-fold subdivision into 'inarticulates' and 'articulates', with an emphasis on the presence or absence of articulatory structures along the hinge, endured (e.g. Carlson 1991a). An alternative phylogenetic model, developed by Gorjansky & Popov (1985), proposed a bi-phyletic origin with the lingulates (brachiopods with organophosphatic shells) placed apart from other brachiopod groups (calcareous-shelled 'inarticulates' and'articulates'). Popov et al. (1993) andHolmer et al. (1995) indicated that separation of organophosphatic and calcareous shell types does not necessarily require a polyphyletic origin for the brachiopod body plan (but see e.g. Valentine 1977; Wright 1979 for an alternative view and Carlson 1995 for discussion); this was supported by phylogenetic analyses that remarkably recalled the earlier brachiopod phylogenetic analysis published in the pioneering book by Hennig (1966). These data were assimilated and reworked in a later phylogenetic analysis by Williams et al. (1996), with the recognition of three brachiopod subphyla, namely Linguliformea, Craniiformea and Rhynchonelliformea; craniiforms and rhynchonelliforms were considered to be sister taxa and the enigmatic chileates as basal rhynchonelliforms (see also discussion in Carlson 2007).

Implications of molecular data
Traditionally brachiopods, together with phoronids and bryozoans (Lophophorata sensu Hyman 1959), were considered to be the sister group of the Deuterostomia. However, with progress in phylogenetic molecular studies, most analyses currently place lophophorate phyla close to the trochozoan clades, firmly establishing the monophyly of the Lophotrochozoa (for more details see Telford 2006;Nesnidal et al. 2014). As discussed below, however, the precise phylogenetic position of the various lophophorate clades, including the brachiopods, remains controversial.
The most comprehensive (ribosomally-based) analyses of lophophorate and brachiopod molecular phylogeny, in terms of the number of species sequenced, have been published by Cohen (2000Cohen ( , 2013 and Cohen & Weydmann (2005). In their earlier analyses, trees were rooted in either the chiton Acanthopleura (Cohen et al. 1998;Santagata & Cohen 2009)  Clathrina (Cohen 2000); their results consistently placed phoronids as an ingroup within the brachiopods. Cohen (2013) favoured a molecular clock-based rooting of potential ingroup taxa. The outcome also positions phoronids as an ingroup within brachiopods, while craniiforms and linguliforms combined emerge as a sister group of the phoronids. Moreover, discinids appeared to form a sister group of the craniides, which is in strong contradiction with available morphological and ontogenetic data. Sperling et al. (2011), however, correctly pointed out that alternative morphological and molecular (ribosomally-based) phylogenies have similar topologies, but differ in rooting; they tested the monophyly of brachiopods using a combined set of seven nuclear housekeeping genes plus three ribosomal subunits (5.8S, 18S, 28S rDNA) as well as analysing the distribution of specific-microRNA (miRNA) genes. The combined analysis (Sperling et al. 2011, fig. 2A) supported brachiopods and phoronids as sister groups, with the monophyly of brachiopods moderately supported (posterior probability of 0.82) and placed craniiforms at the base of the brachiopod clade. It was also noted that analyses of 'homogeneous sites only' and 'heterogeneous sites only' datasets resulted in different topologies and position of the root, with the latter supporting brachiopod paraphyly with phoronids appearing as a sister-group of craniides. Moreover, Sperling et al. (2011) demonstrated that the brachiopod-specific microRNAs favoured brachiopod monophyly, and their absence in phoronids cannot be readily explained by loss of morphological characters. This conclusion, however, was challenged by Cohen (2013, p. 89). Regarding brachiopod interrelationships, Sperling et al. (2011) grouped craniiforms and linguliforms into a single cluster, but as discinids were not included in their analyses, the result remains inconclusive. Nevertheless, Sperling et al. (2011) acknowledged that in 'molecular phylogenetic analyses, the craniides behave as a 'rogue taxon' with little statistical resolution at the base of Brachiopoda'.
In the outcomes of the analysis of nuclear-encoded housekeeping genes presented by Helmkampf et al. (2008, figs 2, 3), brachiopods, represented by craniiforms and rhynchonelliforms, are placed at the base of lophotrochozoan clade, as a sister group of all other lophotrochozoans; bryozoans and phoronids share a common ancestry with annelids and molluscs. Thus, lophophorates appear to be paraphyletic. Data on linguliforms (lingulids and discinids) were not part of that analysis; it is thus of little help in resolving phylogenetic relationships within the Brachiopoda.
The analyses by Nesnidal et al. (2013Nesnidal et al. ( , 2014) have re-introduced support for the Lophophorata as a monophyletic clade, and moreover suggested sister-group relationships with bryozoans and phoronids. Nesnidal et al. (2013Nesnidal et al. ( , 2014 also supported Lophophorata as a sister group of the Nemertea, together forming a sister group to the Annelida. They are thus derived in relation to molluscs and imply that lophophorates evolved from ancestors with a trochophoran larva. Nesnidal et al. (2013Nesnidal et al. ( , 2014 have also claimed that brachiopods are distinctly separate from phoronids, in contrast to other phylogenetic analyses (e.g. Cohen & Weydmann 2005; Cohen 2013). However, the linguliform taxa (e.g. extant lingulids and discinids) were not included. Thus, while the clear separation of phoronids from craniiforms and rhynchonelliform brachiopods is justified, the phylogenetic relation of that group to linguliforms remained unresolved in the molecular phylogeny presented by Nesnidal et al. (2014).
Investigation of the role of hox gene clusters (Schiemann et al. 2017) suggests that the hard tissues of the annelids, brachiopods and molluscs (i.e. the chaetae and shells) share a common origin dating back to the early Cambrian. This implies that Wiwaxia belongs to the lophotrochozoan stem (see also Smith 2014), although the latter taxon possesses distinctly molluscan characters. In these analyses, it is significant that the phoronids are presented as a sister group to the phylum Brachiopoda.
In summary, molecular data show good support for lophotrochozoan monophyly, with the identification of some key novelties, but neither the position of the brachiopods in relation to other phyla within the lophotrochozoan clade, nor the interrelationships between the three major brachiopod subphyla can be resolved with any certainty. The current state of flux in lophotrochozoan molecular phylogeny was predicted by Nielsen (2002), who concluded that based on the 18S rDNA gene sequences, only the branch leading to the Deuterostomia in strict sense can be recognized; the sequence of older speciation events within lophotrochozoan phyla cannot be convincingly resolved at present.

Possible brachiopod stem and sister groups
Although data from embryology and comparative anatomy have been used in support of lophophorate polyphyly, with brachiopods placed as a sister-group of the deuterostomes (L€ uter 2001;Nielsen 2002), rapidly accumulating molecular data (discussed above) now firmly support the position of brachiopods within the lophotrochozoan clade (Fig. 1B). Nevertheless, the monophyletic nature of the Lophophorata, which has been placed in doubt by some earlier studies (e.g. Halanych et al. 1995;L€ uter 2001;Paps et al. 2009), finds some support in published studies on molecular phylogeny (e.g. Nesnidal et al. 2013Nesnidal et al. , 2014 and more traditional comparative anatomic studies (Temereva & Tsitrin 2015). It implies that the lophophores in brachiopods, phoronids and bryozoans are homologous, and probably the only recognizable synapomorphy for the group. Moreover, new evidence from exceptionally-preserved specimens from the middle Cambrian Burgess Shale and Spence Shale suggests that the hyoliths too may be members of this clade (Moysiuk et al. 2017). However, the phylogenetic relationships of brachiopods with ectoproct bryozoans remain controversial (Nielsen 2001(Nielsen , 2002(Nielsen , 2005. The phoronids have frequently been considered to be a sister group of brachiopods (e.  Ushatinskaya 2008). Lingulosacculus may on the other hand represent a 'pre-mineralized' shell condition in the lingulellotretides, but it is more likely that the shell of Lingulosacculus was 'loosely' mineralized, as manifest in mickwitziids and some siphonotretides (Skovsted & Holmer 2003;Holmer & Caron 2006). The earliest lingulate assemblages from the Cambrian Stage 3 show significant morphological diversity, suggesting earlier divergence, which probably occurred before development of a strongly phosphatized shell (Popov 1992, p. 419; Ushatinskaya 2008).
Thus, the position of brachiopods within the lophotrochozoan clade cannot be determined with any accuracy (see Fig. 1B). With the exception of phoronids, there is no other lophotrochozoan group that can be nominated as a stable outgroup for phylogenetic analyses; moreover, the position of phoronids in relation to brachiopods remains controversial. Nielsen (2005, p. 439) noted that the trochophore larva represents an important character, which most probably evolved only once; it is also supported by cell-lineage studies (Nielsen 2012). This indicates that brachiopods and phoronids together form sister groups at the base of all other lophotrochozoan phyla. Another common feature of the phylum is the radial cleavage pattern, which may confirm the close relationship between brachiopods and phoronids or represent a plesiomorphic state. Nielsen (2002, p. 44) stated that 'although traces of spiral pattern have been reported by a few authors it is now agreed that cleavage is radial'. Thus, 'secondary' references to the presence of a spiral cleavage pattern in phoronids (e.g. Nesnidal et al. 2013) are probably erroneous.

THE FIRST BRACHIOPODS
Can we predict the morphology and life mode of the first brachiopod? There is a view that the sessile mode of life in the lophophorates is a secondary adaptation which evolved from a vagile, 'slug-like' life style (e.g. Nesnidal et al. 2014); but this view is not conclusively supported by either the general anatomical and morphological organization of the animal (i.e. having a lophophore rather than a foot) or by the available data from the palaeontological record. In fact, a 'tubular' sessile life habit may be primitive within the lophotrochozoans (e.g. Zhang et al. 2013Zhang et al. , 2014Zhuravlev et al. 2015). Strong palaeontological support for this hypothesis is provided by the uniquely well-preserved tubular fossil Yuganotheca Zhang, Li & Holmer in Zhang et al., 2014 from the Cambrian Stage 3 Chengjiang Lagerst€ atte (Yunnan, China) that exhibits an unusual combination of phoronid and brachiopod characters, notably a pair of agglutinated valves, enclosing a horseshoe-shaped lophophore, supported by a lower bipartite tubular attachment structure with a long pedicle (Zhang et al. 2014). Yuganotheca was placed phylogenetically as a sister group to organophosphatic taxa (Fig. 2). Apart from indicating the rooting of brachiopods into the sessile lophotrochozoan Yuganotheca, this taxon also suggests that the origination of the brachiopod twin-shelled Bauplan preceded the biomineralization of bivalved shells in crowngroup brachiopods (Zhang et al. 2014). The lophophore anatomy of Yuganotheca also indicates that it may not have had a real brachiopod-like, laminar filter-feeding organ, directed laterally through the lophophore as in all living brachiopods, but rather had a more phoronid-like filtration system with the current directed towards the mouth (Fig. 3). The more expansive hypothesis of the relationships between crown-group brachiopods and the tommotiids ( evidence. Notably, the presence of exceptionally-preserved phosphatized setae within the setal canals of the tommotiid Micrina (Butler 2015) has now been demonstrated. These setal structures are identical to the setal canals and exceptionally preserved setae in mickwitziid brachiopods, including Mickwitzia and Heliomedusa from the lower Cambrian of Baltica and China, respectively, and the shell structure of the mickwitziids also includes columns that are closely similar to the columnar shell structure of both Micrina and more derived The view that a sessile life style may be primitive for the lophophorates is further supported by the fact that the gymnolaemate bryozoans, as well as rhynchonelliform and craniiform brachiopods, completely lack a muscular body wall, while the lingulates, phylactolaemate bryozoans as well as phoronids have well developed dermal muscles (Hyman 1959;Holmer et al. 1995;Schwaha & Wanninger 2012). If we assume that the absence of dermal muscles is a plesiomorphic feature, while the body wall musculature evolved convergently in the derived groups of those phyla, an ancestry of the lophophorates from vagile lophotrochozoans looks extremely unlikely. As convincingly demonstrated by Nielsen (2005), it is most likely that the trochophore larva evolved just once and may represent a key synapomorphy within lophotrochozoans. Thus the secondary loss of this larval type in the lophotrochozoans (Nesnidal et al. 2014), secondary loss of the dermal muscles in selected groups of bryozoans and gymnolaemates, and loss of the trochozoan coelom (involving the 4d-cell and a spiral cleavage; L€ uter 2000) appear extremely unlikely. Instead we strongly support lophophorates as an early lophotrochozoan offshoot, or as a paraphyletic branch from stem-group lophotrochozoans.

EARLY PALAEOZOIC RADIATION OF THE BRACHIOPODA
Three major events dominated the diversity and evolution of the Early Palaeozoic Brachiopoda: The Cambrian Explosion, the Great Ordovician Biodiversification Event and the end Ordovician Extinction.

The Cambrian explosion and Cambrian Evolutionary Fauna
Cambrian faunas were dominated by a range of nonarticulated groups, together with groups of disparate articulated taxa, such as the chileates, naukatides, obolellides, kutorginides, billingsellides, protorthides (Fig. 4), orthides and pentamerides (Fig. 5). These groups participated in a variety of loosely-structured, nearshore palaeocommunities but with a clear partition between shallow-water carbonate and siliclastic environments characterized by higher proportions of rhynchonelliforms, and deeperwater finer-grained deposits, often dysoxic, commonly with linguliforms. Key evidence has been extracted from some of the early-mid Cambrian Lagerst€ atten, preserving not only a diversity of form but also exquisite anatomical features (see below). By the early Cambrian, the phylum had already evolved a spectrum of life styles (Topper et al. 2015) exploiting its diversity across a variety of ecological niches. In fact, most of the key life modes had been established prior to the Great Ordovician Biodiversification Event (Topper et al. in press). Challenges, however, exist in understanding the relationships between the individual groups near the base of the stem, particularly in clarifying the relationships between and within the tommotiids.
Origin and early history of the linguliforms Subphylum Linguliformea, as presently defined (Williams et al. 1996), includes all brachiopods with organophosphatic mineralized shells, and is subdivided into two classes. One of these, the short-lived Paterinata, includes the oldest known brachiopods, which appeared at the base of the Cambrian Stage 2 (Aldanocyathus sunnaginicus Biozone) in Siberia (Pelman et al. 1992) and vanished during the Hirnantian Mass Extinction Event at the end of the Ordovician (Harper et al. 2014). The second, Lingulata, appeared late in Cambrian Stage 2 and can still be found in Recent seas. The linguliforms represent an important component of the Cambrian Evolutionary Fauna and, by the beginning of the Ordovician, they show remarkable ecological expansion extending from near-shore to basinal environments. They often occur in benthic assemblages from marine marginal environments such as eutrophic basins, mobile sands in shore-face zones, and as pioneers of the abyssal depths associated with hexactinellide sponges and pterobranchs (Bassett et al. 1999;Tolmacheva et al. 2004).
While separation of lingulates from craniiforms and rhynchonelliforms is robust, paterinates display a mosaic combination of characters, also typical of chileates and rhynchonellates (Williams et al. 1996(Williams et al. , 1998Holmer et al. 2011). The group is probably polyphyletic, rooted in different taxa of the stem group brachiopods. The lingulates share features such as a canaliculated condition of the stratiform shell, a mantle permeated by the intermedial and marginal vascula terminalia, gonads confined exclusively to the body cavity, outer mantle lobes without lobate cells, whereas a single subenteric gangliation is plesiomorphic for the clade (Holmer et al. 1995;Williams et al. 1996). A U-shaped alimentary canal with the anteriorly placed anus and a pedicle developed as the ventral mantle outgrowth with the extension of the coelomic cavity as a core is present even in the earliest lingulates from the Chengjiang Konservat Lagerst€ atte (Zhang & Holmer 2013), although they have a muscular body wall unlike rhynchonelliforms and craniiforms; however, these features are also present in the phoronids and may represent plesiomorphic characters evolved in some stemgroup brachiopods. The paterinates, unlike lingulates, are characterized by a mantle canal system including gonad sacs and with exclusively marginal vascula terminalia, grouped posteromedially-located adductor muscle scars in both valves, and a strophic shell probably fused along the hinge by the strip of periostracum. All these features also occur in the rhynchonelliforms (Williams et al. 1996(Williams et al. , 1998. The earliest lingulates exhibit considerable morphological disparity (Bassett et al. 1999); they had already acquired a conveyor-belt system for shell secretion, which probably evolved at the 'pre-mineralized' state. By contrast, the earliest paterinates (e.g. Askepasma, Pelmanotreta and Solanygolina) exhibit a simple 'stacking' pattern of mineralized shell secretion, probably, predating the origin of the conveyor-belt system, which first became evident only within the family Paterinidae (Williams et al. 1998), although the canaliculated condition, characteristic of lingulates, did not evolve within the paterinates.

The Great Ordovician Biodiversification Event
During the Ordovician, the craniiforms diversified with the craniopsides and trimerellides appearing for the first time. Within the rhynchonelliforms, two major clades, the rhynchonellates and strophomenates, have been identified within the broad-frame classification of the phylum (Williams et al. 1996). Two types of dentition, the simple deltidiodont and the more complex cyrtomatodont, are both phylogenetically and ecologically significant (Jaanusson 1971). Within the two rhynchonelliform subclasses, a laminar secondary shell layer characterizes many of the strophomenates whereas a fibrous secondary layer typifies the rhynchonellates. During the Ordovician radiation, the deltidiodont orthides and strophomenides dominated faunas. Many taxa were first generated around Early-Middle Ordovician island complexes (Neuman 1984;Bruton & Harper 1985) and later dominated the platforms, where they participated in an offshore movement of palaeocommunities (Rong et al. 1999;Bassett et al. 2002) and the occupation of carbonate mudmound and reef structures (Harper et al. 2004;Harper 2006); the latter environments became progressively occupied by the cyrtomatodont athyridides, atrypides and rhynchonellides. By the end of the Ordovician the majority of shell morphologies, excluding perhaps oyster-like forms exemplified by the bizarre Permian lyttonioids, had evolved occupying a wide variety of niches on the seabed. The expansion of the subphylum was evident with increased a-diversity as more species were packed into communities, b-diversity as communities expanded offshore and into carbonate structures and c-diversity as the fragmentation of provinces, particularly during the Early-Middle Ordovician (Harper et al. 2013), drove allopatric speciation (Harper 2006(Harper , 2010. Following the end Ordovician extinction event, spire-bearing brachiopods reached their dominance, particularly in the carbonate environments of the mid-Palaeozoic (Harper & Rong 2001); the strophomenates lost their dominance.

Origin and early history of the craniiforms
The Craniiformea is a small, but distinctive group of calcareous-shelled nonarticulated brachiopods, which emerged early in the Ordovician as a minor component of the Palaeozoic Evolutionary Fauna (Popov et al. 2013a). They reached maximum morphological disparity in the Late Ordovician and Silurian, where they are represented by three orders: Craniida, Craniopsida and Trimerellida ( Fig. 5A-C). Only the craniides survived through all five major extinctions and can be found in modern oceans. While the monophyly of craniiforms is presently well defined (Gorjansky & Popov 1985, 1986Holmer et al. 1995;Popov et al. 1996Popov et al. , 2007a, their relationship to other major brachiopod clades remains controversial. Recent attempts to root craniiforms in some Cambrian taxa have failed. The enigmatic Heliomedusa from the early Cambrian Chengjiang Lagerst€ atte, Yunnan Province, earlier thought to be the ancestral craniopside . This shell-opening mechanism differs markedly from that found in the lingulates, which have well developed dermal muscles (Hyman 1959;Popov et al. 1993;Holmer et al. 1995). Trimerellides are the only craniiforms which developed an effective hinge mechanism (Gorjansky & Popov 1985, 1986 and are also among the largest Early Palaeozoic brachiopods. The craniiforms lack a pedicle attachment during all ontogenetic stages in which an encrusting or ambitopic (initially attached but later free-living) life mode is adopted.
The alimentary canal of craniiforms is axial, supported by complete ventral and dorsal mesenteria, and terminated with an anus placed posteromedially on the posterior body wall. They have paired subenteric ganglia. Some extinct craniiforms (trimerellides) developed rudimentary articulation and a direct shell-opening mechanism, using diductor muscles evolved from the internal oblique muscles (Gorjansky & Popov 1985, 1986. Some features, such as mantle canals containing gonads, a peripheral arrangement of vascula terminalia and lecithotrophic larvae without shells, are shared by the craniiforms and rhynchonelliforms (Holmer et al. 1995;Williams et al. 1996). Mantle lobes in the craniiforms developed without reversion and the appearance of the ventral mantle lobe is significantly delayed. The larva in craniiforms is lecithotrophic with mineralized shell secretion occurring only after settlement, while secretion of the ventral valve is delayed until the end of metamorphosis (Popov et al. dorsal valve and organophosphatic shell mineralization occur in the paterinides. Nevertheless, it is unlikely that the Salanygolinidae represent a transitional form between the organophosphatic paterinates and calcareous chileates (Fig. 4H, J-N). The primitive character of a densely laminate shell structure, which is characterized by a stacking honeycomb pattern with individual units, was probably enclosed by the organic membranes. Similar shell structure is also known in the tommotiids (Balthasar et al. 2009) and it clearly suggests the absence of the conveyorbelt system of shell secretion characteristic of other brachiopods. Thus, phosphatic shell mineralization may have evolved independently, within stem group brachiopods at the base of the craniiform clade. Although the Class Chileata may not belong to the basal rhynchonelliform clade as presented in previous phylogenetic analyses (e.g. Williams et al. 1996) it may represent a stem group craniiform, probably linked with the tommotiides (Holmer et al. 2009). Craniides may have evolved as paedomorphic chileates, retaining larval attachment by the retardation of the secretion of mineralized ventral valve and acquiring holoperipheral growth of both valves. It is also likely that calcareous mineralization of the shell in craniiforms evolved independently from that in rhynchonelliforms. An aragonitic shell is confined exclusively to the craniiform clade, being inferred for the chileate families, Isogrammidae and Trifissuridae (Fig. 4M)   typical 'modern' brachiopods and diversified, substantially, during the Ordovician (Fig. 6). In the earliest taxa, the pedicle probably emerged through the delthyrial notch or delthyrium, but a minute apical foramen is reported in some groups that is of doubtful function. The pedicle, together with the type of interactions between the valves and the development later of lophophore supporting structures, were critical in evolving new taxa and life styles within the class. Some of the earliest rhynchonelliforms occur in the upper Tommotian, including the obolellides (Fig. 4A, B, F, G) and kutorginides (Fig. 4P, T). By the Atdabanian the fauna includes nisusiids (Fig. 4C-E, I), again lacking teeth but nevertheless hinged, whilst the earliest of the more typical rhynchonelliform exemplars, the protorthids (Fig. 4W-Y) (1996), are fairly robust in the broadest sense with the establishment of the rhynchonellate and strophomenate clades (Figs 4,5) in the Cambrian while additional traits, mainly focused on lophophore supports, sequentially define new groups throughout the Ordovician. Thus, crown group taxa such as the Lingulida appeared deep in the Cambrian, and were joined by the craniides (Fig. 5A-C) and by more diverse and dominant rhynchonellide (Fig. 5D-U) faunas later in the Ordovician. Thus, many of the key body plans were already in place by the Cambrian, but in terms of an escalation of families, genera and species, the Ordovician was critical (Fig. 6;Harper & Drachen 2010;Harper et al. 2015). The two main clades, the rhynchonellates and strophomenates, presented alternative life modes; both had deltidiodont dentition, cardinal areas and simple brachial supports but the former taxa were mainly pedunculate whereas the latter were mainly ambitopic or recumbent, taking advantage of both hard and soft substrates. The two clades also differ in their broad   biogeographical distribution, with pentamerides ( Fig. 5I, J, N, O, R, S) generally dominating in lower latitudes and the orthides (Fig. 5F, H) and strophomenides (Fig. 5D, G) widespread but commonly diverse in higher latitudes (Harper et al. 2013). In the stem of the strophomenates, an understanding of the billingsellides (Fig. 4O, R; Fig. 5E) is critical in deciphering the origins of the polytoechiids (Fig. 4V) and clitambonitides (Popov et al. 2001;Topper et al. 2013) together with the position and role of Arctohedra (Fig. 4Q, S, U) and its relationship to the protorthides and clitambonitides. The diversity of the strophomenate clade accelerated during the mid-Ordovician with the expansion of the Plectambonitoidea (Fig. 5D) and in the later Ordovician, the Strophomenoidea (Fig. 5G) (Cocks & Rong 1989;Rong & Cocks 1994;Cocks & Rong 2000). Recent phylogenetic analyses of parts of the group (e.g. Candela 2011a, b;Congreve et al. 2015) have provided more clarity to the classification and evolution of the strophomenides during the Ordovician and Silurian, but questions remain regarding the placement of a number of groups such as the toquimiids, that apparently possess orthoid characters. The cladistic classification of the orthidines (Williams & Harper 2000) remains relatively robust, but that of the dalmanellidines (Harper 2000) is more fragile, with the suggestion that punctuation in that group may be polyphyletic (Benedetto & Muñoz 2017). This, together with the addition of much new morphological data from taxa near the base of the clade since publication of the Treatise, offers the prospect of a better understanding of this complex and currently, poorly-resolved group.

End Ordovician extinction
This extinction, the first major such event affecting animal-based communities, is one of the 'big three' in taxonomic terms (Bambach 2006). It appears not to have been particularly taxon selective, targeting deep-water and warm-water communities (Finnegan et al. 2016) and generating a large number of Lazarus taxa (Rong et al. 2006). Importantly, though, it had a relatively mild impact on the marine ecosystem (Bambach et al. 2004;Bambach 2006;Harper et al. 2014). In terms of the four levels of ecological impacts of extinction crises (see Droser et al. 2000) only third-and fourth-level palaeoecological changes were triggered during the end Ordovician mass extinction, invoking only community or community-type changes during the event. The ecological severity of the event was deemed even less significant than that of the Serpukhovian (McGhee et al. 2012) and is currently ranked only sixth within the eleven largest Phanerozoic ecological crises (McGhee et al. 2013). The two-phased extinction, nevertheless, provided the first real test of the resilience and sustainability of brachiopods, tipping the balance in favour of more derived rhychonelliform morphologies, such as those of the atrypides (Fig. 5K, Q), athyridides (Fig. 5L, M), pentamerides and spiriferides (Fig. 5U) (Harper & Rong 2001;Huang et al. 2017).

THE BROAD-FRAME CLASSIFICATION: CHALLENGES MOVING FORWARD
There have been no substantive attempts to reassess the phylogeny of the Brachiopoda in its entirety since the landmark study of Williams et al. (1996;Fig. 7). Attempts prior to this, with the exception of the cladistics analysis of Carlson (1991a) focused on several key characters (e.g. Williams 1956;Williams & Rowell 1965;Rudwick 1970;Williams & Hurst 1977) rather than total evidence. Nevertheless, new data, new investigative techniques and the more precise location of fossil data in time and space provide the opportunity to test existing phylogenetic hypotheses and suggest alternatives. Morphological data remain crucial. New data, particularly from those groups originating in the Cambrian, provide some exciting challenges to conventional wisdom. We focus on some of the recent research (see also above) pertinent to any substantive revision of the broad-frame classification of the phylum.
We have briefly indicated six key areas that merit discussion and exploration: 1. The paterinates as a natural outgroup for the phylum.
Morphological and stratigraphic data suggest this group offers to be the most appropriate outgroup for phylogenetic rooting. 2. The significance of the chileates. This group is increasingly important for understanding initial divergence of craniiform, strophomenate and rhynchonellate clades, as well as for early evolution of the attachment structures in the brachiopods. They include the oldest known brachiopods with a calcareous and strophic shell, and an unusual pedicle emerging through a vertical umbonal penetration; a colleplax, typical of the chileates, is also present in the phosphatic Salanygolina and this structure may have equivalents in the craniiforms and strophomenides. The group thus demonstrates a puzzling mosaic of characters developed elsewhere in apparently more distantly related taxa. 3. Mutual relationships amongst the kutorginides, naukatides and obollelides and with the strophomenates. New morphological and stratigraphical data on the first three groups requires a re-evaluation of their relationships with each other and the strophomenates. The attachment structures of the kutorginates and strophomenates are probably homologous and quite distinct from those of the rhynchonellates. Investigation of the early ontogenetic stages of the kutorginides is key to resolving their relationship with the strophomenates. 4. The position of the pentamerides within the rhynchonelliformeans. Pentameride dentition has been considered intermediate between cyrtomatodont and deltidiodont modes or to include both types. While many pentamerides are astrophic, the cyrtomatodont condition has not been established with any veracity. Moreover, the presence of platforms in both valves might suggest a lack of muscles with tendons in contrast to those of the orthides. On this basis it would seem unlikely that the pentamerides are members of the brachiopod crown group. 5. The composition and mutual relationships of the three great clades. The three subphyla, the Linguliformea, Craniiformea and Rhynchonelliformea were clearly separated by the Early Ordovician (possibly earlier) on the basis of their shell structures and compositions together with their respective morphologies. Thus, there is a definite possibility that all three major brachiopod lineages were phylogenetically distinct, prior to mineralization of their respective shells. 6. Evolution and timing of shell mineralization. The origin of the phylum and its earliest evolution is associated with the development of different types of secretory mechanisms and regimes together with the utilization of different shell substances. Understanding the respective origins and evolutionary trajectories of the different types of mineralization is still in its infancy but the new many new taxa, recognized in the brachiopod stem, show considerable prospect for unravelling this complex problem.