Exceptionally prolonged tooth formation in elasmosaurid plesiosaurians

Elasmosaurid plesiosaurians were globally prolific marine reptiles that dominated the Mesozoic seas for over 70 million years. Their iconic body-plan incorporated an exceedingly long neck and small skull equipped with prominent intermeshing ‘fangs’. How this bizarre dental apparatus was employed in feeding is uncertain, but fossilized gut contents indicate a diverse diet of small pelagic vertebrates, cephalopods and epifaunal benthos. Here we report the first plesiosaurian tooth formation rates as a mechanism for servicing the functional dentition. Multiple dentine thin sections were taken through isolated elasmosaurid teeth from the Upper Cretaceous of Sweden. These specimens revealed an average of 950 daily incremental lines of von Ebner, and infer a remarkably protracted tooth formation cycle of about 2–3 years–other polyphyodont amniotes normally take ~1–2 years to form their teeth. Such delayed odontogenesis might reflect differences in crown length and function within an originally uneven tooth array. Indeed, slower replacement periodicity has been found to distinguish larger caniniform teeth in macrophagous pliosaurid plesiosaurians. However, the archetypal sauropterygian dental replacement system likely also imposed constraints via segregation of the developing tooth germs within discrete bony crypts; these partly resorbed to allow maturation of the replacement teeth within the primary alveoli after displacement of the functional crowns. Prolonged dental formation has otherwise been linked to tooth robustness and adaption for vigorous food processing. Conversely, elasmosaurids possessed narrow crowns with an elongate profile that denotes structural fragility. Their apparent predilection for easily subdued prey could thus have minimized this potential for damage, and was perhaps coupled with selective feeding strategies that ecologically optimized elasmosaurids towards more delicate middle trophic level aquatic predation.


Introduction
Plesiosaurians (Plesiosauria) were highly diverse Mesozoic marine amniotes whose fossil record extended over 135 million years. During this vast timeframe the clade achieved a variety of body forms and feeding modes ranging from massive-skulled megacarnivores (e.g. the historically referred to Scanisaurus [31][32][33][34], to reconstruct a prototype dental array. These specimens derived from the Museum of Evolution palaeontological collection (PMU) at Uppsala University in Sweden (Table 1), and were recovered from a stratigraphically restricted horizon of the uppermost lower Campanian (Late Cretaceous:~80 mya) Belemnellocamax mammillatus belemnite zone at Åsen and Ivö Klack-two geographically proximal localities (~6 km apart) in the northeastern Kristianstad Basin of Skåne in southern Sweden [34,35]. Almost all of the PMU Scanisaurus teeth exhibited damage from pre-burial abrasion, surface weathering and/or excavation, which precluded volumetric assessments of mean size based on complete teeth [30,36,37]. However, correlation between the maximum crown length/basal diameter dimensions in our sample (Fig 3) suggested consistent linear proportions despite breakage. The tooth size range (PMU 24468 at 8.1/5.6 mm, to PMU 24533 at 47.3/11.2 mm) also closely matched those recovered from in situ elasmosaurid dentitions (e.g. premaxilla/ maxilla 10-45/5-10 mm; dentary 25-30/~7 mm [19]).  Four teeth from Åsen (PMU 28757-PMU 28760), and three from Ivö Klack (PMU 28761-PMU 28763) were selected for thin sectioning based on overall crown completeness and/or intact roots. These specimens were embedded in bicomponent epoxy resin (Lamit 109, Kittfort) and impregnated with EpoFix (Struers) prior to cutting on a 150 mm diamond blade, and abrasion to 0.2 mm on a Montasupal grinder (grits: 240, 400, 600) using silicon carbide. The resulting 25 mounted petrographic slides (Fig 4) were inspected on a Leitz DM RXE microscope equipped with a Leica DFC 550 camera and LAS v.4.2 software. Diagenetic modification was assessed using Field emission gun scanning electron microscopy (FEG-SEM) undertaken on uncoated slides with a Zeiss Supra 35-VP (Carl Zeiss SMT) incorporating a low vacuum VPSE detector, Robinson BSD backscattered electron-imaging attachment, and coupled EDX Apex 4 (Ametekh) EDS-detector for dispersive X-ray microanalysis. This detected high component phosphate, implying minimal elemental alteration in the samples from Åsen (Fig 5). Extensive mineral recrystalization otherwise entirely obscured dentine microstructures in the teeth from Ivö Klack (Fig 6A and 6B) rendering them unsuitable for further examination.
In accordance with previous studies [30,36,37], we used a continuous longitudinal midline section through the apex (PMU 28757/1) and crown base (PMU 28757/2) of the most complete available tooth (proportionately sized within one standard deviation of the sample mean: Fig 3; Table 1) to calculate tooth formation time. However, other longitudinal (PMU 28757/3, PMU 28757/4) and transverse sections were made through additional teeth (PMU 28759/1-2, PMU 28758/1-5) to inspect enamel/dentine microstructures in different aspects. Dentine deposition rates were based on counts of short-period incremental growth lines of von Ebner [30,[36][37][38], which are laid down following a circadian biorhythm in all tooth-bearing vertebrates [30,[36][37][38][39][40][41][42]. Nevertheless, the identification of such markings has differed between recent studies [37,39], with Dean [40][41][42] providing an explicit characterization of von Ebner lines as narrowly-spaced fine layers <3 μm apart, and distinct from coarser Andresen lines, which can be up to 20 μm and represent longer-cycle periodicity. This pattern of two types of interpolating increments was visible in our fossils, therefore we proceeded by taking measurements based on high-resolution digital micrographs in ImageJ [43,44]. Sequential sets of 10 clearly visible short-period lines ( Table 2) were demarcated by the leading edge of one light-  Tooth formation in elasmosaurid plesiosaurians coloured increment band to the next, and measured at both labial and lingual locations perpendicular to the enamel-dentine junction, as well as from different locations closer to the cuspal apex (where dentine formation rates are known to increase [40]) and below the mid-length of the crown. Distal and medial locations were also examined in transverse sections to crossreference results. A cumulative average of all short-period increment measurements was then multiplied by the mean dentine thickness taken across each location (Table 3); this established an estimate of the total dentine accumulation rate for each individual tooth [38]. All petrographic slides (PMU 28757/1-4, PMU 28758/1-5, PMU 28759/1-2, PMU 28760/ 1-3, PMU 28761/1-4, PMU 28762/1-2, PMU 28763/1-5) and accompanying image metadata produced by the described study are publically accessible through the PMU fossil collection and digital archives. Abbreviations: db, discoloured dentine band; dr, diagenetic recrystalization; dt, dentinal tubules; edj, enamel-dentine junction; er, enamel ridge; igz, interglobular zone; md, microstructural degradation; ol, osteocyte lacunae; mb, microboring; pc, pulp cavity.

Results
Microstructural examination of the thin (~120-150 μm) prismatic enamel layers revealed discrete dark-coloured laminations (Fig 6C and 6D). These might represent striae of Retzius, which deposit over variable timeframes during enamel formation in modern amniotes [42,45]. The laminations also became more pronounced within the external enamel ridges ( Fig  6D), perhaps in response to mechanical stress incurred during force loading across the curved lingual surface of the tooth in life. Odontocyte lacunae and radially arrayed dentinal tubules were observed extending from the pulp cavity (Fig 6C and 6E). Longitudinal central tubules projected towards the crown apex ( Fig 6F). Extensive microstructural degradation along the enamel-dentine junction ( Fig  6E, 6G and 6H) resembled microborings attributed to post-depositional bacterial or fungal activity [38,39,46]. Some sections otherwise preserved an interglobular zone at the orthodentine periphery (Fig 6G and 6H), and the tooth roots (where visible) incorporated an external layer of cementum.
Discolouration from secondary minerals highlighted faint concentric staining up to~50 μm in thickness, which was only visible in some transverse sections (Fig 6I). Occasionally this intercalated with widely-spaced dark and light coloured bands up to~30 μm apart (Fig 6J) that could represent traces of long-period Andresen lines accumulated over multiple days [39,40,42]. Finer alternating layers 1.2-5.3 μm wide (Fig 6K; Table 2) corresponded with incremental markings elsewhere interpreted as daily lines of von Ebner [39,40,42]. We therefore adapted the approach of Gren and Lindgren [38] in multiplying the mean line thickness of 1.8-3.2 μm from the longitudinal midline section through PMU 28757/1-2, by the corresponding dentine thickness range of 1750-3000 μm measured at the apex and crown base, to estimate a total of Tooth formation in elasmosaurid plesiosaurians 547-1667 periodic increments across the whole tooth (Table 4). This yielded an average dentine depositional rate of 950 days based on the cumulative mean of all measurement locations [38], and an extraordinarily prolonged tooth formation time of 2.6 years (full range = 1.5-4.5 years from the apex versus tooth base) following approximation of the Cretaceous year at 370.3-371 days [47,48] (Table 4).

Discussion
Our evidence for protracted tooth formation in dental remains attributed to the Late Cretaceous elasmosaurid Scanisaurus concurs with previous reports of extended dental replacement cycles in plesiosaurians and other more basal sauropterygians [18,[26][27][28]. Indeed, delayed replacement periodicity might have been broadly characteristic of these clades (and been influenced by thecodont gomphosis, indeterminate growth and metabolism [49]), but has been more specifically linked to regionalized tooth shape variation and enlargement in heterodont pliosaurids [18]. The detection of prolonged tooth formation times in Scanisaurus could therefore indicate a comparable adaptation towards discretized tooth function, which in anisodont elasmosaurids presumably involved selective elongation of the premaxillary-maxillary and dentary 'fangs' via modified growth periods and rate of odontoblast/ameloblast differentiation during apical dentine secretion [42]. Nevertheless, not all elasmosaurids were anisodont [13,20,21], implying that disparate tooth arrangements could evince contrasting formation rates between elasmosaurid taxa (a phenomenon recognized elsewhere in polyphyodont amniotes [37,50]). This has particular pertinence for Scanisaurus, which is frequently compared with homodont elasmosaurids such as Aristonectes [31,34], but may have alternative affinities based on its delayed tooth formation pattern. Another important implication is the increased likelihood of severe tooth wear, breakage and/or related disease [30,37]. Certainly, dental pathologies have been described in plesiosaurians before (mainly larger-skulled 'pliosauromorphs' [10,51]), as has chronic tooth-associated bone deformation [52]. Notably, densely spaced dentine increments [39], apparently slow tooth replacement [53], and caries [54] have also all been documented in ichthyosaurians. In contrast, only superficial wear and missing teeth have thus far been described in elasmosaurids [6,10], which likewise lacked robust crowns more typically associated with prolonged tooth development and macrophagous diets [10,30,38]. This might be explained by preferential feeding (inferred from bromalite contents [10,14,16]) on relatively small, easily subdued prey that was swallowed whole and processed using gastroliths in the gut [55]. Such dietary specialization potentially evolved hand-in-hand with the need to minimise excessive tooth damage, and was likely further coupled with selective foraging strategies that employed the slender interdigitating teeth to entrap, pierce or sieve prey from the water column and fine seafloor sediments (e.g. silt and mud); non-selective suspension feeding has been critiqued because of structural constraints on the reptilian pharynx [56]. Delayed tooth formation, together with 'alveolarized' replacement and heterogenous dental configurations could therefore have been a key factor constraining the adaptive radiation of elasmosaurids as middle trophic-level aquatic predators, and perhaps indirectly contributed to their environmental [57,58] and geographical prevalence [19] via the capacity to utilize a wider range of available food resources.