Osteohistology of the silesaurid Sacisaurus agudoensis from southern Brazil (Late Triassic) and implications for growth in early dinosaurs

the non-dinosaurian dinosauriform silesaurids are the closest relatives of crown-group dinosaurs and are thus, important for understanding the origins of that group. here, we describe the limb bone histology of the Late triassic silesaurid Sacisaurus agudoensis from the Candelária Sequence of the Santa Maria Supersequence, Rio Grande do Sul state, Brazil. The sampled bones comprise eight femora and one fibula from different individuals. The microscopic analysis of all elements reveals uninterrupted fibrolamellar bone tissue indicating rapid growth. A transition to slower growing peripheral parallel-fibered bone tissue in some individuals indicates a decrease in growth rate, suggesting ontogenetic variation within the sample. the osteohistology of Sacisaurus agudoensis is similar to that of other silesaurids and supports previous hypotheses that rapid growth was attained early in the dinosauromorph lineage. however, silesaurids lack the complex vascular arrangements seen in saurischian dinosaurs. Instead, they exhibit predominantly longitudinally-oriented primary osteons with few or no anastomoses, similar to those of some small early ornithischian dinosaurs. this simpler vascular pattern is common to all silesaurids studied to date and indicates relatively slower growth rates compared to most Dinosauria.

Given the discovery of new silesaurid specimens (Kammerer et al. 2012, Peecook et al. 2013, Barrett et al. 2015) and the current debate regarding their phylogenetic relationships to Dinosauria, data on the life history of these animals can provide important new information on their biology and shed light on their similarities/ differences to Dinosauria. the analysis of bone microstructure or osteohistology of fossil bones can provide such life history data by revealing growth patterns and rates, ontogenetic stages and even biomechanical and hence lifestyle adaptations of extinct animals (e.g., Enlow and Brown 1956, 1957, 1958, ricqlès 1976, horner et al. 2000, Chinsamy-turan 2012, Padian and Lamm 2013, Legendre et al. 2016. Previous studies on the osteohistology of extant archosaurs and their extinct relatives have shown differences in the type of bone tissue deposition between the two lineages of the group. Within the Pseudosuchia, lamellar bone, indicative of slow growth, is preponderant (ricqlès et al. 2003). In contrast, the bone tissues of the Ornithodira reveal predominantly fibrolamellar bone (Padian et al. 2001), which is defined here by the presence of a woven-fibered bone matrix associated with primary osteons, a tissue type that is indicative of relatively high rates of bone deposition and growth rates (Francillon-Vieillot et al. 1990). the presence of growth marks [annuli and Lines of Arrested Growth (LAGs) indicating a temporary decrease or cessation in growth, respectively, Francillon-Vieillot et al. 1990] varies considerably in both groups, but most pseudosuchians tend to exhibit cyclical growth from early ontogenetic stages (ricqlès et al. 2003) whereas many derived ornithodirans only express growth marks during the latest ontogenetic stages (e.g., neosauropod dinosaurs, Klein and Sander 2008).
Studies on the osteohistology of dinosaurs have increased dramatically in recent years (e.g., Grady et al. 2014, Vanderven et al. 2014, Cerda et al. 2014, Bo et al. 2016, Mitchell et al. 2017, Skutschas et al. 2017. however, given the relative scarcity of material, fewer studies encompassing the taxa leading to crown-group dinosaurs have been conducted. Previous descriptions on silesaurid osteohistology include Silesaurus opolensis Dzik 2003 (Fostowicz-Frelik and Sulej 2010), Asilisaurus kongwe Nesbitt et al. 2010 (Griffin andNesbitt 2016) and Lewisuchus admixtus romer 1972 (Marsà et al. 2017). these studies reveal similar bone tissues amongst the studied taxa, where uninterrupted fibrolamellar bone with isolated primary osteons dominates. Griffin and Nesbitt (2016) suggested that the absence of LAGs may be widespread amongst silesaurids and Marsà et al. (2017) further proposed that these similarities indicated that silesaurids shared similar life history strategies.
here, we present the first osteohistological description of the Late triassic Brazilian silesaurid Sacisaurus agudoensis Ferigolo and Langer 2007. Given that few studies on the osteohistology of silesaurids have been conducted to date, the data obtained from S. agudoensis provides an important An Acad Bras Cienc (2019) 91(Suppl. 2) e20180643 3 | 17 contribution towards understanding the presumed similarities in their life histories and more broadly, the acquisition of early dinosaur osteohistological characteristics and growth patterns.  Langer 2007, Langer andFerigolo 2013) assigned to the upper strata of the Candelária sequence and the Riograndia Assemblage Zone, Norian in age (Soares et al. 2011, Langer et al. 2018. the lithology of this sedimentary package mostly consists of sandstones with isolated layers of mudstone and mud intraclasts, which is associated with fluvial channel deposits (Zerfass et al. 2003(Zerfass et al. , horn et al. 2014. Sacisaurus agudoensis is a small-bodied species (less than 1 meter in body length), with elongated distal hind limbs, an herbivorous/ omnivorous diet and well-developed cursorial abilities (see Langer and Ferigolo 2013). the bone accumulation of S. agudoensis shows a peculiar preservation of over 30 nearly identical right femora with only one left femur (see Langer and Ferigolo 2013). the morphology of the taxon has already been described in detail in previous studies (see Langer 2007, Langer andFerigolo 2013).

INStItUtIONAL
We analyzed nine limb bones, which comprise eight right femora and one right incomplete fibula, all from different individuals. The complete femora in the bone accumulation range from 85 mm to 110 mm in length (Langer and Ferigolo 2013). We sectioned one nearly complete femur (85 mm length, without the head) and seven incomplete femora, where total length could not be measured (table   I)   transverse thin sections were made as close to the midshaft as possible as this region undergoes the least secondary remodeling during growth and thus, provides the most complete growth record of the animal in long bones (Chinsamy 1990, Francillon-Vieillot et al. 1990, Chinsamy and Dodson 1995, horner et al. 1999 they were cut into smaller blocks perpendicular to the long axis of the bone using a cut-off diamondtipped saw within a Ken 9025 grinding machine. One surface of each resin block was then affixed to a frosted petrographic glass slide using the same resin that was used for embedding and left to set for a further 24 hours. the sections were wet-ground to approximately 60 µm thick and polished using a Prazis APL-S polishing machine with abrasive papers of increasing grit size (P80, P120, P320, P400, P600, P1200, P1500, P2000, P3000).

IMAGE ANALYSIS
the resulting thin sections were analyzed using a Leica DM 2500P microscope and photographed using an attached AxioCam Erc 5s camera. the image analysis program NIh ImageJ 1.52e (Schneider et al. 2012) was used to quantify the degree of vascularization of the mid-cortex of An Acad Bras Cienc (2019) 91(Suppl. 2) e20180643 5 | 17 each element allowing a quantitative comparison between the study specimens and other related groups in which similar data is available. Channel area was measured in a field of view (FOV) taken from the mid-cortex, divided by the total FOV area and then converted into a percentage (Chinsamy 1993a). As already noted in the literature, although the channels includes lymph, nerves and vascular canals that extend through the bone tissue (Mckenzie and Klein 2000), the calculation provides an estimation of the maximum amount of vascularization within a given section. We repeated the procedure for every third FOV at 10X magnification resulting in eight to ten FOV per slide, providing a good average of a given section. terminology adopted follows that of Francillon-Vieillot et al. (1990), reid (1996, Chinsamy-turan (2005Chinsamy-turan ( , 2012 and Prondvai et al. (2014).

FEMOrA
the transverse thin sections of all femora reveal a medullary cavity free of trabecular bone. Some elements (MCN PV 10022; PV 10062; PV 10228) are lateromedially compressed resulting in a flattened medullary cavity. This compression has caused numerous cracks to extend through the compact bone, but there are enough patches of wellpreserved bone tissue to facilitate an assessment of the primary bone. A thick, irregular avascular endosteal layer of lamellar bone (Figure 1a-b) surrounds the medullary cavity in all femora. the bone matrix exhibits regions of closely distributed haphazardly arranged globular osteocyte lacunae in a woven-fibered bone matrix between primary osteons (Figure 1c-f). this bone matrix indicates static osteogenesis and fast growth (Prondvai et al. 2014) and combined with the primary osteons acknoledges the presence of a fibrolamellar bone complex. the primary osteons are predominantly isolated longitudinally-oriented canals (Figure 2a), with some short radial and circumferential anastomoses ( Figure  2b). the vascularization of the mid-cortex varies from 10.6% to 14.5% (table I). Specimen MCN PV 10070 exhibits a more complex vascular arrangement, especially in the inner cortex, with radial and circumferential anastomoses reaching a sub-reticular arrangement (Figure 2c-d). Small, scattered secondary osteons were observed in most of the bones, but are mostly limited to the inner cortex. however, the specimens, MCN PV 10230, MCN PV 10231 and most notably MCN PV 10071, exhibit large, abundant secondary osteons, but not to the extent of forming multi-generational dense haversian bone. In MCN PV 10230 and PV 10231 (Figure 3a and 3b, respectively) they appear locally, on the medial side of the bone, whereas in MCN PV 10071 (Figure 3c-d) they extend all around the medullary cavity and into the mid-cortex, occupying approximately half of the compact cortex. this region forms compacted coarse cancellous bone and is demarcated by a reversal line (Figure 3d). the high degree of remodeling in these three bones reflects the region in which the thin sections were taken, which are the proximal (MCN PV 10230 and PV 10231) and distal (MCN PV 10071) metaphyses where remodeling tends to be more prevalent.
Growth marks are absent from all the bones, resulting in an uninterrupted fibrolamellar bone complex. however, a peripheral area of slower forming parallel-fibered bone can be seen in some of the femora (MCN PV 10022; PV 10062; PV 10070; PV 10230; PV 10234). In these specimens, the osteocyte lacunae near the bone periphery are flattened and more evenly distributed, running parallel to one other (Figure 4a-h). the region also appears anisotropic in polarized light (Figure 4af). In these elements, there is a slight decrease in the size and amount of the vascular canals towards the peripheral region of the bone. the decreased vascularization and transition to a parallel-fibered   bone matrix in the outer regions of the bone indicates an overall decrease in the growth rate. however, the absence of an External Fundamental System (EFS, sensu Cormack 1987) suggests that these bones were not fully grown as an EFS generally indicates the achievement of maximum size.

FIBULA
An incomplete fibula (MCN PV 10084, Figure  5a-d) was thin sectioned for this study. the most striking feature of the bone is a broad and rugose "tibial flange" extending obliquely from its cranioproximal corner (see Langer and Ferigolo 2013: Figure 19f). Similar to the femora, the medullary cavity of the fibula is completely free of trabecular bone ( Figure 5a). the compact cortex also comprises a fibrolamellar bone complex with longitudinallyoriented primary osteons in a woven bone matrix (Figure 5c). the primary osteons are mostly isolated with short anastomoses and the bone tissues are less vascularized (8.1%) than the femora (average = 12.2%). Vascularization remains constant throughout the cortex and there are no interruptions in the bone tissue (Figure 5d). Scattered secondary osteons are mostly limited to the inner cortex. As in the femora an EFS is absent. the most conspicuous difference between the bone tissues of the fibula and femora is the absence of an endosteal layer surrounding the medullary cavity in the fibula (Figure 5a and 5b).

ONtOGENEtIC StAGE OF thE StUDIED INDIVIDUALS
the analysis of the bone tissues of the study material reveals similar osteohistological features and thus suggests that the elements represent a similar ontogenetic stage. All elements, apart from the fibula, exhibit a thick avascular endosteal layer of lamellar bone surrounding the medullary cavity. this structure, as well as an EFS, has been observed in the femora of mature individuals of the silesaurid Silesaurus opolensis (Fostowicz-Frelik and Sulej 2010), indicating the cessation of medullar expansion and thus, a mature ontogenetic age (Chinsamy-turan 2005). however, in the case of S. agudoensis, the absence of an EFS suggests that the circumferential endosteal lamellae indicates a temporary cessation in medullary expansion, after which growth would have resumed. the presence of fibrolamellar bone, even at the sub-periosteal surface in some elements, indicates that these animals were still growing at the time of death. Some femora (MCN PV 10022; PV 10062; PV 10070; PV 10230; PV 10234) exhibit a transition from an inner woven bone matrix to an outer parallel-fibered bone matrix. This transition from fast to slower growing bone, as well the decrease in size and abundance of vascular canals towards the bone periphery is evidence of a decrease in growth rate, which indicates a departure from the juvenile stage into at least an early subadult stage for these individuals.   The transverse thin section of the fibula MCN PV 10084 lacks all features indicating an older ontogenetic stage with only scattered secondary osteons appearing mainly in the inner cortex. there is no decrease in vascularization and no change to slower forming parallel-fibered bone towards the outer surface. Although vascularization is lower than in the femora, the uniformity of the rapidly forming fibrolamellar bone throughout the cortex supports a juvenile status for this individual.

GrOWth PAttErN AND COMPArISON WIth
OthEr ArChOSAUrOMOrPhA the bone tissues of Sacisaurus agudoensis are primarily composed of uninterrupted fibrolamellar bone. the presence of this tissue type indicates relatively high rates of bone deposition and hence growth (Amprino 1947, Francillon-Vieillot et al. 1990, Margerie et al. 2002, Padian and Lamm 2013, Prondvai et al. 2014. the compact bone comprises isolated longitudinally-oriented primary osteons with some short anastomoses and a high density of globular osteocyte lacunae. Fibrolamellar bone has been found in all other silesaurids studied to date and the vascular pattern in S. agudoensis is similar to that observed in other taxa, such as Silesaurus opolensis (Fostowicz-Frelik and Sulej 2010), Asilisaurus kongwe (Griffin and Nesbitt 2016) and Lewisuchus admixtus (Marsà et al. 2017). However, it differs slightly from S. opolensis and L.admixtus in having more anastomoses connecting the longitudinal primary osteons. Given the onset of peripheral parallel-fibered bone in some individuals of S. agudoensis and absence of growth marks in all the bones studied, it is clear that it grew continuously (Padian and Lamm 2013) to at least the subadult stage of its ontogeny. Growth marks have been observed within peripheral parallel-fibered bone in the largest specimen of Silesaurus opolensis, however, (Fostowicz-Frelik and Sulej 2010) and thus, it is possible that S. agudoensis exhibited cyclical growth during later ontogenetic stages, but material from older individuals is required to test this.
the growth patterns of silesaurids differ from that seen in stem archosaurs, i.e., nonarchosaurian archosauromorphs ( Figure 6). For example, the osteohistology of the Late triassic Trilophosaurus buettneri Case 1928, described by Werning and Irmis (unpublished data), comprises poorly vascularized lamellar bone with multiple LAGs throughout the cortex, indicating a slow growth rate. rhynchosauria osteohistology also indicates slow growth rates. For example, Werning and Nesbitt (2015) described the osteohistology of a femur and tibia from a mature individual of the Middle triassic Stenaulorhynchus stockleyi haughton 1932 from tanzania as moderate to poorly vascularized parallel-fibered bone tissue. Similar bone tissues have also been found in the ribs of Hyperodapedon sanjuanensis Schultz 2000 (previously Scaphonyx sanjuanensis Sill 1970) (Mukherjee 2015). However, the fibrolamellar bone in these taxa is restricted to the innermost cortex, the rest of the cortex being comprised of slowly forming lamellarzonal bone (Mukherjee 2015, Veiga et al. 2015. the tibia from an adult individual of the slightly more derived Early triassic archosauromorph Prolacerta broomi Parrington 1935, exhibits a vascular pattern similar to that found in silesaurids in which the longitudinally-oriented primary osteons are mostly isolated with a few anastomoses (Botha-Brink and Smith 2011). however, the bone tissues are notably less vascularized (e.g. 2.2% compared to an average of 11% in S. agudoensis ) and combined with the prevalence of a mixture of weakly developed fibrolamellar and parallelfibered bone, these features suggest lower growth rates in P. broomi (Botha-Brink and Smith 2011).
Fibrolamellar bone becomes more prevalent in the non-archosaurian archosauriforms. Proterosuchus fergusi Broom 1903, a mediumsized proterosuchid (nearly 2 m in body length, Botha-Brink and Smith 2011), exhibits highly vascularized (e.g. femoral vascularity 15%) radiating vascular canals within fibrolamellar bone during early ontogeny. Growth decreases dramatically during late ontogeny, however, with the onset of poorly vascularized lamellar-zonal bone (Botha-Brink and Smith 2011). An unidentified long bone of the proterochampsid Chanaresuchus bonapartei romer 1971 reveals similar bone tissues, but with a more sub-reticular vascular arrangement in the inner cortex (ricqlès et al. 2008). Several elements from different individuals of the large-bodied triassic erythrosuchids Erythrosuchus africanus Broom 1905 (Gross 1934, ricqlès 1976, ricqlès et al. 2008, Botha-Brink and Smith 2011 and Garjainia madiba Gower et al. 2014(Gower et al. 2014) exhibit a variety of vascular arrangements including laminar, reticular and radial, within a fibrolamellar bone complex, indicating rapid growth rates, similar to those found in fast-growing dinosaurs. Although faster growing tissues become more prevalent among non-archosaurian archosauriforms, it should be   (Nesbitt et al. 2009). these taxa show that there was some degree of experimentation amongst the triassic non-archosaurian archosauriforms as shown by the variety of bone tissue patterns in this group. however, the increasing prevalence of highly vascularized fibrolamellar bone shows that rapid growth rates evolved before the origin of Ornithodira (ricqlès et al. 2008, Botha-Brink and Smith 2011, Gower et al. 2014) and is considered plesiomorphic for archosauriforms (ricqlès et al. 2003). the simpler vascular pattern seen in silesaurids differs from most early dinosaurs where the vascular arrangements tend to exhibit more complex anastomosing. For example, plexiform and laminar vascular patterns are typical of early saurischian dinosaurs [e.g., Herrerasaurus ischigualastensis reig 1963, (Padian et al. 2001, 2004, ricqlès et al. 2003 (Klein and Sander 2007, Cerda et al. 2014, Chinsamy, 1993b] and the coelophysid neotheropod (Chinsamy 1990, ricqlès et al. 2003, Padian et al. 2004) Megapnosaurus rhodesiensis Ivie et al. 2001(previously Syntarsus rhodesiensis raath 1969, as well as the early ornithischian Lesothosaurus diagnosticus Galton 1978(specimen NMQr 3076, Knoll et al. 2010. In contrast, the osteohistology of some early ornithischians such as the heterodontosaurid Fruitadens haagarorum Butler et al. 2010 and the thyreophoran Scutellosaurus lawleri Colbert 1981 exhibit slower forming bone tissues. F. haagarorum exhibits parallel-fibered bone (Butler et al. 2010) and S. lawleri contains very little fibrolamellar bone during early ontogeny and entirely lamellarzonal bone during adulthood (Padian et al. 2004). Both taxa contain predominantly longitudinallyoriented vascular canals with few anastomoses. these ornithischians are relatively small, with F. haagarorum only growing to a maximum length of 75 cm and S. lawleri less than 1.5 m (Padian et al. 2004, Butler et al. 2010) and thus, body size may have played a role in their relatively slower growth rates (Padian et al. 2004). For example, Padian et al. (2004) showed that the vascular arrangements of the bone tissues in small dinosaurs are simpler and less vascularized compared to their larger relatives (Singh et al. 1974, Case 1978. however, the fabrosaurid L. diagnosticus, was also relatively small with an approximate body length of 2 m (Knoll et al. 2010). It displays rapidly forming fibrolamellar bone with longitudinally-oriented primary osteons during early ontogeny and a laminar vascular arrangement during mid-ontogeny (Knoll et al. 2010), typical of other early saurischian dinosaurs. thus, factors other than body size likely played a role in shaping ornithischian growth rates and more research is required to better understand the life histories of these taxa.
Although there are relatively few studies on the osteohistology of silesaurids and early dinosaurs, the data available to date suggest that a simpler vascular arrangement of longitudinallyoriented primary osteons is typical of silesaurids. In contrast, most early dinosaurs examined to date [apart from the contentious Nyasasaurus parringtoni ] show a more complex vascular arrangement early in their life history. Given that more complex vascular patterns are associated with higher growth rates (Margerie et al. 2002, Padian andLamm 2013), this implies that silesaurids grew relatively more slowly than the earliest crown-group dinosaurs (apart from some small ornithischians). Again, this may be related to  Barrett et al. 2015), all silesaurids studied to date were small-bodied species (body length of approximately 1.5 m or less, Barrett et al. 2015), whereas early sauropodomorph dinosaurs with more complex anastomoses were larger (Irmis 2011, turner and. however, as noted with L. diagnosticus some taxa do not show correlation between body size and vascularization. the coelophysid neotheropod M. rhodesiensis also exhibits highly vascularized, complex arrangements despite having a similar body size to S. opolensis (Padian et al. 2001, Irmis 2011, turner and Nesbitt 2013. Additionally, Barrett et al. (2015) described the first larger-bodied silesaurid (NHMUK R16303) from the Lifua Member of the Manda Beds of tanzania, estimated to have had a femoral length of approximately 345 mm (body length estimated to be approximately 3 m), which according to the authors, may represent a large individual of Asilisaurus kongwe. the osteohistology of A. kongwe indicates that the specimens were still growing at the time of death and thus probably represent juveniles or subadult individuals (Griffin and Nesbitt 2016). the material we sampled here also indicates that the bones came from juveniles and subadult individuals that were still growing. this suggests that some silesaurids reached larger body sizes than previously thought. Further osteohistological descriptions of such large specimens will clarify the correlation between body size and vascular pattern among silesaurids.

CONCLUSIONS
the bone tissues of the non-dinosaurian dinosauriform silesaurid S. agudoensis exhibit rapidly forming fibrolamellar bone with longitudinally-oriented primary osteons and occasional anastomoses similar to that observed in other silesaurids. Our study supports previous studies in showing that high growth rates were already present prior the evolution of Ornithodira. however, the simpler vascular pattern of longitudinally-oriented primary osteons with few anastomoses differs from the complex arrangements generally found in crown-group dinosaurs. this suggests that silesaurids grew comparatively more slowly than most dinosaurs, which may be related to phylogeny or smaller body sizes. Further examination of more dinosauromorph taxa is required to confirm this hypothesis.

ACKNOWLEDGMENTS
We thank Dr. C.S. Vega from the Universidade Federal do Paraná (UFPr) and the Laboratório de Pesquisa em Microscopia of UFPr for access to imaging equipment. We also thank two anonymous reviewers for their suggestions and comments. the Fundação de Amparo à Pesquisa do Estado do rio Grande do Sul (FAPErGS) provided financial support to FHV, MBS and JBB (Grant number 0441-2551/14-7), the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) provided financial support to FHV, the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) provided financial support to MBS (Grant number 312387/2016-4), to AMr (Grant number 306951/2017-7) and the National research Foundation (UID 98819 and 104688), the Palaeontological Scientific trust (PASt), Johannesburg, South Africa, and DSt-NrF Centre of Excellence in Palaeosciences (CoE-Pal) provided financial support to JBB.

AUTHOR CONTRIBUTIONS
Fábio hiratsuka Veiga, Jennifer Botha-Brink and Marina Bento Soares are responsible for the study conception, data analysis and to the writing of the manuscript with input from the other authors. Fábio hiratsuka Veiga, Ana Maria ribeiro and Jorge