Among living vertebrates, only the gnathostomes (the clade of living jawed vertebrates) possess a mineralized skeleton. The gnathostome clade comprises chondrichthyans (sharks, rays, and holocephalans) and osteichthyans (all other living jawed vertebrates). Naturally, therefore, chondrichthyans have been pivotal in attempts to understand skeletal evolution among osteichthyans (Williamson, 1849, 1851; Hertwig, 1874a, b, 1876, 1879, 1882; Stensiö, 1961; Ørvig, 1951, 1968, 1977; Reif, 1976, 1978a, b, 1980, 1982). To be sure, chondrichthyans are a natural outgroup to osteichthyans, but they do not necessarily reflect an ancestral gnathostome state. This expectation is true of the chondrichthyan skeleton, and most especially of the micromeric dermal skeleton, which is reduced with respect to the primitive gnathostome condition, as evidenced by successive extinct sister lineages to the chondrichthyans, osteichthyans and, indeed, to crown gnathostomes (Donoghue and Sansom, 2002). These lineages include the jawed acanthodians and placoderms, and the jawless ostracoderms (osteostracans, galeaspids, heterostracans, thelodonts, and anaspids), all of which possessed an extensively developed mineralized dermal skeleton composed of bone and, in many lineages, dermal tubercles comprising enameloid and dentine (Donoghue and Sansom, 2002). With respect to the histological structure of the dermal skeleton in the gnathostome crown ancestor, the most poorly known and yet most significant of these extinct lineages are the placoderms, the evolutionary relationships of which remain the focus of vigorous debate (Fig. 1; e.g., Johanson 2002; Brazeau, 2009; Young, 2010; Davis et al., 2012). Perceived as a clade, Placodermi has been most widely considered a sister lineage to crown-gnathostomes (Fig. 1A; Young, 2008, 2010). Perceived as an evolutionary grade, placoderms comprise a series of successive sister-lineages to crown-gnathostomes (Fig. 1B; Johanson, 2002; Brazeau, 2009; Davis et al., 2012). In either instance, placoderms remain integral to understanding the histological structure of the dermal skeleton in the ancestral crown gnathostome, the condition from which both chondrichthyans and osteichthyans are derived. Thus, we set out to survey the histology of the dermoskeleton across placoderm diversity. In providing an insight into the tissues present in the placoderms, we attempted to resolve the plesiomorphic condition of the gnathostome dermal skeleton and, in consequence, provide insight into the subsequent evolution of this skeletal system in extant vertebrates.
Synthesis of Previous Research into the Histological Structure of the Placoderm Dermal Skeleton
Research into the microstructure of placoderm dermal armour has been ongoing since the 19th century (e.g., Agassiz, 1844; Gürich, 1891). However, there has been no systematic attempt to survey skeletal composition and reports are disparate both in their taxonomic and anatomical coverage, with little comprehensive understanding of how tissues and structures are distributed across different placoderm groups. Here, we review the sum of published knowledge.
The dermal skeleton is perceived generally to consist of a superficial compact lamellar layer (referred to as the “Oberflächenschicht” by Heintz, 1929, or, within phyllolepids, as the “Skulpturschicht” by Gross, 1934), a cancellous spongiosa, and a compact lamellar base. The presence of distinct layers and their relative thickness varies from one account to another; in Phyllolepis, Stensiö (1934) notes that the superficial layer is confined to the ornament, and Hills (1931) identifies a multilayered basal layer. Heintz (1929) recognized two distinct layers of spongiosa within the dermal armour of Heterogaspis (Monaspis; contra Downs and Donoghue, 2009): a superficial-most “Maschenschicht” with irregular interconnected cavities, and a basal-most “Kanalschicht” with cavities typically parallel to the basal surface. The spongiosa is also purported to contain Haversian canals (Bystrow, 1957). Superficial tubercles are widespread in arthrodires. These may be composed of bone or semidentine, the latter of which is a putative synapomorphy of placoderms (Goujet and Young, 2004; Young, 2010) defined by its unipolar cell lacunae. Semidentine cell lacunae were first identified in placoderms by Gross (1935) and were initially referred to as “Unipolare Knochenzellen” (unipolar bone cells), reclassified as the tissue “semidentine” by Ørvig (1951). Ørvig's reinterpretation was not initially accepted universally and the term was rejected by some authors, including Bystrow (1957) who preferred to interpret the tissue as bone, and Kulczycki (1957) who thought it transitional between bone and dentine. Enamel has not been found in the arthrodires; the thin layer of enamel identified in Sedowichthys by Bystrow (1957) has been widely disregarded, and Stensiö's (1925) enamel-coated “coccosteid” jaw has been reinterpreted as acanthodian (Miles and Young, 1977). Bystrow (1957) identified an apparent evolutionary transition within arthrodires, from an ancestral state of thin plates bearing “dermal teeth” (with dentine and enamel) to an increasing thickness of plates and filling in of “teeth” with bone. Stensiö (1934) identified Haversian canals in the basal layer of Phyllolepis.
Antiarchs also exhibit a three-layered structure, with a superficial lamellar layer (termed the “Tuberkelschicht” by Gross, 1931), a cancellous spongiosa, and a compact basal lamellar layer (Grundlamellenschicht after Gross, 1931). This structure is typified by Asterolepis ornata (Heintz, 1929). Juvenile specimens of A. ornata and Bothriolepis canadensis exhibit three layers even in early ontogeny (Gross, 1931; Downs and Donoghue, 2009), with the compact layers being the first to develop. Antiarchs differ from Arthrodira in the presence of a basal cancellous component in the superficial layer. In addition, the superficial layer as a whole is separated from the middle spongiosa by a “planar discontinuity” (Goodrich, 1909; Gross, 1935). This was interpreted by Downs and Donoghue (2009) as a cementing line, and also as a plane of overlap between adjacent scarf joints. The superficial layer may possess tubercles that, in derived forms at least, are composed of cellular bone. It has been speculated that semidentine is present in ancestral antiarchs (Young, 2008), but this has not been substantiated. Both the superficial and middle layers contain evidence of extensive resorption and secondary bone growth (Downs and Donoghue, 2009). The presence of spheritic mineralization in all three layers of the dermoskeleton has led to a number of different interpretations. Stensiö (1931) considered the tissue to be an intermediary between bone and cartilage, whereas Ørvig (1968) classified it as “globular bone.” Burrow (2005) suggested the tissue was cartilage and thus part of the endoskeleton. Downs and Donoghue (2009), demonstrated that the topographic position of the spheritic mineralization, between dermal bone layers, is incompatible with an endoskeletal interpretation, instead interpreting the spheritic mineralization as evidence for rapid growth of dermal bone.
The dermal skeleton has been interpreted as two-layered, consisting of a compact surface layer and inner canal layer (Wells, 1944). Wells' figures contradict this simplistic interpretation, and a basal, possibly lamellar, layer appears to be present (Plate VII, Figs. 2 and 6, Wells, 1944). Compact dentine has been described in Palaeomylus (Wells, 1944), and acellular bone in Palaeomylus and Eczematolepis (Ørvig, 1980).
Acanthothoracid histology is known only from scales of Romundina and Murrindalaspis. These contain an upper double-layer of meso/semidentine, a middle spongiosa, and a lamellar base (Burrow and Turner, 1998). Tubercles might overgrow in three generations (Ørvig, 1975).
The histological description is incomplete. The surface layer in Macropetalichthys agassizi is made up of multiple generations of dentine tubercles (Gross, 1935), although this is not referred to by Wells (1944) in his investigation of Macropetalichthys rapheidolabis. In contrast, the tubercles of the primitive Lunaspis are made up of bone (Bystrow, 1957; at the time Lunaspis was considered an arthrodire). There is agreement that the spongiosa contains open and interconnected canals, but only Gross (1935) has figured a basal layer, which he noted is thin and often missing.
Of the rhenanids, only the microstructure of Ohioaspis tumulosa scales has been investigated. Gross (1973) describes three layers of tissue: an “isopedinous” basal layer with possible evidence of bone reworking, a middle layer of vascular canals and previous generations of tubercles, and a “Skulpturschicht” of orthodentine-like semidentine.