Paleohistology of the crocodyliform Mariliasuchus amarali Carvalho & Bertini, 1999 (Mesoeucrocodylia, Notosuchia) based on a new specimen from the Upper Cretaceous of Brazil

Mariliasuchus amarali Carvalho & Bertini, 1999 was a terrestrial quadruped crocodyliform from the Late Cretaceous of the Bauru Group, Brazil. In this paper we present the ﬁ rst study of the bone histology of this species. Moderate growth rate, interspersed by growth marks, was observed in thin sections of a rib and appendicular bones; growth patterns observed in M. amarali s appear similar to those observed in Triassic archosauriforms. Th e M. amarali cross-sections indicate growth variability between both the axial and appendicular bones. Distinctive remodeling processes were found in the radius, which had extensive inner cortex remodeling composed of compacted coarse cancellous bone. Furthermore, the medullary region was inﬁ lled with spongy bone. a été observé dans M. amaralis variabilité de croissance entre les axiaux et appendiculaires. processus de remodelage été trouvés dans le radius remodelage du interne


INTRODUCTION
Archosauromorpha, a group of diapsids that are more closely related to birds and crocodylians than to squamates, has one of the most diverse fossil records within the Sauropsida. It radiated during the Triassic and gave rise to the crown group, Archosauria. Archosaurs diverged into two lineages, the Pseudosuchia (Crocodyliformes and relatives) and the Ornithodira (pterosaurs and dinosaurs, including birds) (Nesbitt 2011;Ezcurra 2016).
Th e Notosuchia were a diverse clade of crocodyliforms that thrived during the Cretaceous and whose remarkable diversity is mainly represented by the Neuquén Basin, Argentina (e.g. Pol & Leardi 2015;Leardi et al. 2018) and the Bauru Group, Brazil (e.g. Marinho & Carvalho 2009;Campos et al. 2011;Pol et al. 2014). Our knowledge of the early members of the Mesoeucrocodylia (sensu stricto Benton & Clark 1988) has been largely derived from notosuchian records in Africa and South America (Ortega et al. 2000). In Brazil, the Bauru Group has yielded a diverse notosuchian fauna, with more than 30 species reported (Iori & Campos 2016;Martinelli et al. 2018).
Mariliasuchus Carvalho & Bertini, 1999 belongs to the advanced notosuchians, a group of heterodont crocodyliforms found in the South American landmasses during the Cretaceous (Pol et al. 2014). Th e genus consisted of two species: Mariliasuchus amarali Carvalho & Bertini, 1999and Mariliasuchus robustus Nobre, Carvalho, Vasconcellos & Nava, 2007, both from Marília County, Adamantina Formation, Bauru Group (Brusatte et al. 2017. Th is formation is a stratigraphic unit that dates from the Campanian-Maastrichtian and its paleoenvironment is classifi ed as a warm semiarid system (Zaher et al. 2006;Andrade & Bertini 2008;Fernandes & Ribeiro 2015). Mariliasuchus amarali was a small-bodied notosuchian (Augusta & Zaher 2019) from this ecosystem. Th e species is considered terrestrial due to the presence of a short skull, lateral orbits, frontal external nares, and long, robust limbs indicating a quadrupedal posture (Vasconcellos & Carvalho 2005). Mariliasuchus amarali is characterized by anatomical features, such as the presence of four premaxillary teeth and fore-aft jaw movements (Pol et al. 2014 et al. 2017). In this study, we describe the osteohistology of the Archosauromorpha using the bones of Mariliasuchus amarali to compare growth strategies and lifestyle to other archosauromorphs taxa.

GEOLOGICAL SETTING
Mariliasuchus amarali UFRPE 5311 remains were collected in an outcrop of the Adamantina Formation on the right margin of the Água Formosa Creek, 10 km south of the Marília County. It is located 500 m from the secondary road between Marília and Ocauçu counties in the state of São Paulo, Brazil (Fernandes & Coimbra 2000;Brusatte et al. 2017). Th e age of this formation is uncertain, however, has been dated to late Campanian-early Maastrichtian ages (Santucci & Arruda-Campos 2011;Brusatte et al. 2017;Batezelli et al. 2018). Th e outcrops cover a broad range of the western part of the São Paulo and Minas Gerais states. histology of this species. Moderate growth rate, interspersed by growth marks, was observed in thin sections of a rib and appendicular bones; growth patterns observed in M. amaralis appear similar to those observed in Triassic archosauriforms. Th e M. amarali cross-sections indicate growth variability between both the axial and appendicular bones. Distinctive remodeling processes were found in the radius, which had extensive inner cortex remodeling composed of compacted coarse cancellous bone. Furthermore, the medullary region was infi lled with spongy bone.

RÉSUMÉ
Paléohistologie du crocodyliforme Mariliasuchus amarali Carvalho & Bertini, 1999 (Mesoeucrocodylia, Notosuchia)  Th is formation overlaps the basalts of the Serra Geral Group which represents a signifi cant volcanic event consequential of the separation of South America and Africa during the Early Cretaceous (Renne et al. 1992). Th e Adamantina Formation presents at its base reddish mudstone and sandstone layers intercalated on planar-parallel bedding. Th e desert lithologies decrease upwards through the formation and the lacustrine and fl uvial clays and massive sandstones begin to appear in the middle-to-upper portion of the unit deposited in warm and humid conditions, with carbonatic nodules and root marks (Marsola et al. 2016;Brusatte et al. 2017). Th e specimen herein was found in a semi-arid paleoenvironment with a fl uvial distributary system.

MATERIAL
UFRPE 5311 is composed of the humerus, ulna, and radius from the right forelimb and three ribs (Fig. 1). Th ese elements were assigned to Mariliasuchus amarali by association with other complete specimens that were diagnosed as M. amarali using cranial and post-cranial morphology. Th e specimen is housed in the paleontological collection at the Universidade Federal Rural de Pernambuco (UFRPE), Brazil. Osteohistological termino logy follows Francillon-Vieillot et al. (1990) and Enlow (1969).

SAMPLING METHODS
Samples of 0.5 cm in thickness were removed from the diaphysis of each element, in order to prepare the histological slides. Th in sections were prepared at the Laboratório de Paleobiologia e Microestrutura, Centro Acadêmico de Vitória of the Universidade Federal de Pernambuco (CAV/UFPE) and at the Laboratório de Paleontologia e Sistemática of the Universidade Federal Rural de Pernambuco (LAPASI/UFRPE).
Th e specimens were hand-measured and photographed as per protocols proposed by Lamm (2013). Th in sections were then prepared using standard fossil histology techniques (Chinsamy & Raath 1992; Lamm 2013). Samples were embedded in clear epoxy resin Resapol T-208, catalyzed with Butanox M50, and cut with a diamond-tipped blade mounted on a saw. Th e mounting-side of the sections was wet-ground using a metallographic polishing machine (Aropol-E, ArotecLtda) with Arotec abrasive sandpapers of increasing grit size (60/ P60, 120/P120, 320/P400, 1200/ P2500) until a 60 μm thick section was reached.

IMAGE ANALYSIS
Th in sections were observed under normal light and crosspolarized light with lambda compensator using two optical microscopes. Images were obtained from an AxioCam digital sight camera (Zeiss Inc., Barcelona, Spain) mounted to an Axio Imager.M2 (Zeiss Inc. Barcelona, Spain) and an Olympus BX51 (Olympus Corporation, Tokyo, Japan), mounted to an Olympus DP26 (Olympus Corporation, Tokyo, Japan). Th e images were taken at CAV/UFPE and Laboratório de Gemo-logia (LABGEM/UFPE). Cortical thickness was measured using straight-line tool on ImageJ. Th e fi nal value was given by the arithmetic mean of the local thicknesses. Th ey were taken from 10 distinct points of each thin section.
Bone compactness analysis was carried out for all bone elements. Th e photographs of the cross-sections were transformed into binary images in Adobe Photoshop® CS6 by marking bone in black and vascular spaces (medullary cavity, vascular canals, and resorption cavities) in white. Th e binary images (supplementary material) were analyzed quantitatively with the software Bone Profi ler version 4.5.8 (Girondot & Laurin 2003) to infer on bone compactness profi le from cross-sections.

RIB
Some remodeling processes in the perimedullary and medullary regions were observed, forming cancellous bone which is highlighted by trabeculae and resorption cavities ( Fig. 2A, B). Cross-sections have a high level of bone compactness (0.89) with the cortical thickness of approximately 0.9 mm. Th e cortex is composed of parallel-fi bered bone tissue interrupted by four single lines of arrested growth (LAGs) and one double LAG (Fig. 2B). Th ese growth marks become closely spaced toward the periosteal margin (Fig. 2C, E) and these lines divide the cortex into fi ve distinct growth zones, the third being the largest (Fig. 2E). Th e periosteal cortex shows simple vascular canals and primary osteons, but scattered secondary osteons are also present (Fig. 2D).
HUMERUS Th e medullary cavity was infi lled with iron oxides during the fossil diagenetic processes (Fig. 3A), also as this bone is fragmented compactness (0.58) must be underrated.
Most of the endosteal region shows primary bone with no trace of remodeling (Fig. 3D). Th e primary bone (cortical thickness approximately 5 mm) appears to be a combination of lamellar and parallel-fi bered bone tissue (Fig. 3B). Th e parallel-fi bered bone tissue is highly vascularized and shows longitudinally oriented primary osteons and primary reticular canals branch up to the external bone surface (Fig. 3E).
In the mid-cortex, at least three LAGs are observed (Fig. 3F). Th e osteocyte lacunae exhibit either fl at or globular shapes that are randomly spread out. Towards the outer margin, some Sharpey's fi bers and four LAGs are visible in the outer cortex (Fig. 3C).
ULNA Th e ulna contains iron oxides and some cracks derived from taphonomic processes (Fig. 4A). Bone remodeling is visible in the endosteal layer (Fig. 4B, C) and the cortical thickness is approximately 4 mm with a bone compactness of 0.88. Th is cortex exhibits a poorly organized parallelfi bered arrangement and contains at least fi ve LAGs and the woven bone matrix is visible in some portions (Fig. 4D,  E). Th e vascularization pattern is composed of longitu-  (17) dinally-oriented vascular canals (Fig. 4D) which become radially-oriented toward the external bone surface (Fig. 4E).
In the lamellar matrix osteocyte lacunae have a fl at aspect distributed in concentric rows.

RADIUS
Remarkably this radius, found in articulation with the previously described humerus and ulna, exhibited a sharply diff erent pattern in bone microstructure. Medullary cavity in the radius is infi lled with spongy bone (Fig. 5A) and the cancellous bone is composed of resorption cavities surrounded by thin lamellar bone tissue (Fig. 5B). In the endosteal region, compacted coarse cancellous bone (CCCB) is constituted of sinuous convolutions of lamellae (Fig. 5C). A scalloped and conspicuous resorption line marks the boundary between the compacted coarse cancellous bone and the periosteal cortex (cortical thickness approximately 1 mm; bone compactness is 0.87) (Fig. 5B). Th e primary lamellar bone tissue is poorly vascularized and growth marks are featured by fi ve LAGs and one double LAG (Fig. 5B, E). Parts of mid-cortex are comprised of woven bone tissue (Fig. 5F). Th is bone contains fl attened and scarce osteocyte lacunae that follow the orientation of collagen fi bers.

COMPARING GROWTH PATTERNS AND LIFESTYLE IN MARILIASUCHUS AMARALI TO OTHER ARCHOSAUROMORPHS
Since the Late Permian, diff erent lineages of Archosauromorpha exhibit very diverse growth patterns, and in archosauromorphs, bone growth appears to be related to posture (Werning & Irmis 2010;Ponce et al. 2017), lifestyle (Woodward et al. 2011;Company & Pereda-Suberbiola 2017;Andrade et al. 2018) or even to both (Ricqlès et al. 2003(Ricqlès et al. , 2008Botha-Brink & Smith 2011;Ezcurra et al. 2014;Mukherjee 2015;Werning & Nesbitt 2016). In early ontogenetic stages, they appear to have higher growth rates, which seems to decrease over time, a pattern that is commonly observed in Permo-Triassic terrestrial archosauromorphs (Allen 2003;Ricqlès et al. 2003Ricqlès et al. , 2008Botha-Brink & Smith 2011;Ezcurra et al. 2014;Mukherjee 2015;Werning & Nesbitt 2016). On the other hand, terrestrial archosauriforms from the Triassic-Jurassic show continuous bone apposition, as evidenced by cortices composed of uninterrupted fi brolamellar tissue (Ferigolo & Langer 2007;Ricqlès et al. 2008;Knoll et al. 2010;Grinham et al. 2019;Marsà et al. 2019;Veiga et al. 2019). Slower growth rates as indicated by cortices composed of lamellar-zonal bone tissue are present in Archosauromorpha at least since the Permian. For example, this is a common growth strategy adopted by extinct and extant crocodyliforms with diff erent lifestyles. Nevertheless, semi-aquatic crocodyliforms are capable of rapid growth for some periods (Table 1). Th e bony elements of Mariliasuchus amarali showed diff erences in the types of the tissue deposited throughout life, a fact considered to be directly related to endogenous responses to biomechanical forces and lifestyle. In the humerus and ulna, growth was still active shortly before death, as indicated by vascular canals being open to the surface. However, the rib presents a slow growth that is evidenced by the poorly vascularized parallel-fi bered bone. Th e bony tissues of the humerus and ulna of M. amarali have similarities to Triassic quadruped archosauriforms such as the proterochampsian Chanaresuchus Romer, 1971(Ricqlès et al. 2008Trotteyn et al. 2013;Ponce et al. 2017;Arcucci et al. 2019;Grinham et al. 2019) and the terrestrial pseudosuchian Batrachotomus kupferzellensis Gower, 1999(Sues & Schoch 2013Klein et al. 2017;Grinham et al. 2019). Th ese species had fast to moderate appositional growth rates with temporary cessation of the bone growth during annual cycles (Ricqlès et al. 2008;Klein et al. 2017).

GROWTH MARKS IN MARILIASUCHUS AMARALI
Both LAGs and annuli are cyclical growth marks (CGMs) driven by circannual changes in the environment (Castanet et al. 1993). However, the LAGs were the only CGMs present in M. amarali and varied in number among the bone elements. Such variability may be explained by the medullary expansion (Woodward et al. 2014) or diff erent rhythms of osteogenesis (Cullen et al. 2014). Th e highest number of LAGs were observed in the UFRPE 5311 humerus, suggesting an age of seven years for this individual at time of death. Th e absence of an external fundamental system (EFS) indicates that growth had not completely ceased in the sampled bone elements, suggesting that skeletal maturity had not been attained by UFRPE 5311. Notwithstanding, the endosteal lamellae surrounding the medullary cavity in the ulna indicate cessation of medullary expansion (Chinsamy et al. 2008). Th ese traits indicate a sub-adult ontogenetic state for this specimen.

BONE REMODELING
Th e large amount of the CCCB in the perimedullary region suggests that the radius is the least resistant within the limb bones, this occurs because primary cortical bone is more resistant than the secondary bones (Ray & Chinsamy 2004). Th e compacted coarse cancellous bone formation occurred during the remodeling process when cancellous bone in the medullary region of the metaphysis was converted into compacted coarse cancellous bone as the metaphysis was relocated and became the diaphysis at an advanced ontogenetic stage (Enlow 1962a, b;Prondvai et al. 2012). Th e presence of this bone in the cortex has been reported in Iberosuchus macrodon Antunes, 1975 (2 m in total length; see Ortega et al. 2000) IPS 4932 in Cubo et al. (2017), and considered by the authors as a compacted spongiosa related to muscle insertion or as a radial fi brolamellar bone.
However, Mariliasuchus amarali (c. 1.4 m body length) shares with the neosuchian Susisuchus anatoceps Salisbury, Molnar, Frey & Willis, 2006 (c. 1.10 m body length) the remodeled trabecular bone in the ribs. In the case of the neosuchian Guarinisuchus munizi Barbosa, Kellner & Viana, 2008 (c. 2.79-3.43 m), the remodeling process in the rib forms Haversian systems (Hastings et al. 2010;Andrade et al. 2015;Sayão et al. 2016). Th e latter must repair microdamages caused by an intense biomechanical strain on the rib (Martin & Burr 1982;Lee et al. 2002). Th e osteohistology of rib described here suggest that the axial skeleton of M. amarali was aff ected by a low strain level, comparable with that seen in Susisuchus anatoceps.

CONCLUSIONS
Mariliasuchus amarali resembles Chanaresuchus (Arcucci et al. 2019) and Batrachotomus kupferzellensis (Klein et al. 2017) in growth patterns, with the presence of woven and parallel-fi bered bone tissues periodically interrupted by growth marks throughout the cortex. Th is suggests that these Triassic quadruped archosauriforms shared similar growth rates. Bone microstructure of Mariliasuchus amarali (UFRPE 5311) indicates active growth.
Th e assessment of intraskeletal variability reveals variable appositional growth within the skeletal elements of Mariliasuchus amarali. Th e compacted coarse cancellous bone in the midshaft region of the radius highlights the hypothesis that this bone had the lowest resistance among limb bones. In the ribs of Mariliasuchus amarali and Susisuchus anatoceps (Sayão et al. 2016), the remodeling process forms trabecular bones, whereas in Guarinisuchus munizi (Andrade et al. 2015) it occurs through Haversian reconstruction.