A skeleton from the Middle Jurassic of Scotland illuminates an earlier origin of large pterosaurs

of the the wingspan statistical analyses. I.B.B. scanned the specimen. G.F.F. conducted the bone histology analysis and photographed the specimen. A.P. discovered the specimen. D.A.R., S.L.B., T.J.C., N.D.L.C., using a 1200-grit silicon carbide abrasive slurry (Kemet International) on a glass plate, and then two 10,000-grit lapping pads (Kemet International), the second of which was impregnated with mineral oil (Johnson & Johnson) to improve optical contrast. observed and using a Leica DMLP Transmitted Light Polarizing Microscope and Leica Application Suite Photomon-tages were using automated ‘photomerge’ tool in Photoshop Creative Cloud Osteocyte density mostly uncrushed and preserved in three dimensions, and it shows no evidence of substantial fracture, compac-tion, or disarticulation, suggesting it was buried in a low-energy environment, which dePolo et al. 11 suggested as biologically active mudﬂats.

In brief Jagielska et al. report a fossil pterosaur skeleton from the Middle Jurassic (ca. 167 million years ago) of Scotland, belonging to a new genus and species, Dearc sgiathanach. With a wingspan of over 2.5 m, it was the size of the largest modern-day flying birds and shows that pterosaurs developed to a large size earlier than previously thought.

Locality and horizon
The specimen was discovered by A.P. in 2017 at Rubha nam Brathairean (Brothers' Point), Trotternish, Isle of Skye, Scotland, in the Lonfearn Member of the Lealt Shale Formation (Bathonian, Middle Jurassic) 9,10 ( Figure S1). The skeleton was embedded in a well-sorted lagoonal bioclastic limestone (rich in Neomiodon, ostracods, and conchostracans), which overlies and infills dinosaur trackways impressed in subaerially exposed mudstones. 11 These units formed in a marginal marine/nearshore environment that fluctuated between submerged and exposed.

Diagnosis
Dearc sgiathanach is a rhamphorhynchine pterosaur with the following autapomorphies: tri-tubular vomers with ''trident-shaped'' precapillary contact, pre-choana depression on the palatal surface of the maxilla, enlarged optic lobes expanded anteroposteriorly, and fourth metatarsal more robust (diameter 2.53) than mt1-3. For additional information, see STAR Methods.
Bone histology and maturity Using Bennett's 12 osteologically based ontogenetic stages for the closely related Rhamphorhynchus, NMS G.2021.6.1-4 has features of terminal-stage adults, such as large and recurved premaxillary teeth, fused scapula-coracoid, well-developed humeral crests, smooth bone texture, and fused long bone epiphyses. However, some osteological features are indicative of immaturity according to Bennett: 12 portions of the skull are unfused, such as the jugal with the lacrimal, and there appears to be limited fusion in the sacral vertebrae.
Immaturity is corroborated by bone histology (STAR Methods; Data S2). The cortex of a sampled wing phalanx is composed entirely of primary fibrolamellar bone 13 and preserves two prominent lines of arrested growth (LAGs), which indicate that the individual was at least 2 years old at death. 14 The position of the second LAG close to the external bone surface suggests that the individual died shortly after emerging from an annual growth hiatus. The cortex is densely vascularized and has a high proportion of woven bone, indicating a rapid rate of growth throughout life. 15 The presence of vasculature extending to the external bone surface and the absence of an external fundamental system indicate that the individual was actively growing when it died. In many respects, the bone microstructure is similar to small, young individuals (<30% adult wingspan; size class I 12 ) of Rhamphorhynchus 16 and other actively growing juvenile pterosaurs, 17,18 indicating that it is best interpreted as a juvenile or subadult 19 that had not reached adult body size when it perished.

Wingspan and body size
Wingspan-defined as double the summed lengths of the bones of a single wing 5 -is tightly correlated to body mass and wing area in pterosaurs and thus a robust body size proxy. 20 A complete wingspan cannot be measured directly from NMS G.2021.6.1-4 because some wing phalanges are missing. To estimate wingspan, we compiled measurements of complete wingspans of two non-monofenestratans represented by large sample sizes-Rhamphorhynchus and Dorygnathus-and regressed these against the lengths of individual bones to create predictor formulas (STAR Methods; Data S1C-S1Q). Using Rhamphorhynchus scaling, the humerus length (112 mm) and skull length (222 mm) of NMS G.2021.6.1-4 indicate wingspans of 3.8 and 2.2 m, respectively. The largest known Rhamphorhynchus (Natural History Museum UK 37002) is considerably smaller, with a wingspan of 1.8 m, humeral length of 79 mm, and skull length of 202 mm. Using Dorygnathus scaling, the humeral length of NMS G.2021.6.1-4 indicates a wingspan of 1.9 m, approximately 10% larger than the largest Dorygnathus (1.69 m wingspan, 84 mm humerus).
These results demonstrate that Dearc is the largest Jurassic pterosaur yet known, consistent with the fact that its humerus and skull are the longest of any Jurassic specimens. Furthermore, we interpret these results as evidence that Dearc likely achieved wingspans over 2.5 m, and perhaps larger (>3 m). This is based on two lines of reasoning. First, we consider the Rhamphorhynchus equations, which give larger wingspan estimates, as the most valid predictors: Rhamphorhynchus is a closer relative of Dearc than is Dorygnathus, is known from a larger sample size (and thus generates a regression with tighter error bars and a higher r 2 value), and has a well-established and nearly isometric growth trajectory that makes predicting wingspan from isolated skeletal elements more justifiable. 21 Second, the holotype of Dearc (NMS G.2021.6.1-4) was an actively growing juvenile-subadult at death and would have been larger as an adult (STAR Methods).
mCT data provide a stellar view of a complete, articulated palate and hyoid of a non-monofenestratan pterosaur in dorsal and ventral view ( Figure 2). The heart-shaped choana is cut medially by forking vomers, comprised of three cylindrical rods that converge anteriorly in a trident-shaped contact. There is a thin extension of the ectopterygoid, which rotates around its own axis, forming an elevated ventral border of the postpalatine fenestra, joining the vomers at a perpendicular angle. This ''contorted'' morphology has not been described in other pterosaurs.
mCT data also provide one of the few brain and inner ear endocasts of a basal pterosaur ( Figure 2). Like Rhamphorhynchus, 7 Dearc had a large cerebrum with optic lobes positioned at the same level as the forebrain and a large flocculus, around which thin and arched semicircular canals looped, that nonetheless did not project to the same lateral level as the cerebrum. In pterodactyloids, however, the brain is highly flexed so that the cerebrum is elevated relative to the optic lobes, and the flocculus is expanded beyond the cerebrum laterally. 7,8,24 In Dearc, the optic lobes are larger, anteroposteriorly longer, and more widely exposed dorsally than in any known basal pterosaur or pterodactyloid (Rhamphorhynchus, 7 Allkauren, 24 and Tapejara 8 ).

Phylogenetic analysis
Our phylogenetic analysis focuses on non-monofenestratan pterosaurs and combines data from several independent published analyses 22,[25][26][27][28][29][30][31][32][33][34][35] with new characters, while excluding taxa known only from highly immature specimens and characters that exhibit strong ontogenetic variation, resulting in a dataset of 58 taxa scored for 155 characters (Data S1B and S2). Dearc sgiathanach is recovered within a large grade of nonmonofenestratans, including subclades centered on Rhamphorhynchus and Scaphognathus ( Figure 4). Dearc is in the former subclade, where it groups with the Chinese Angustinaripterus and Sericipterus, in the clade Angustinaripterini, diagnosed here for the first time by several features including a large antorbital fenestra, reclined quadrate, and proportionally elongate anterior cervicals (see above).  Figure S4 and Data S1 and S2.

Remarks
The holotype of Dearc sgiathanach is a rare three-dimensionally preserved pterosaur from the Jurassic, which gives unique insight into the osteology, size, growth, and neuroanatomy of a basal non-monofenestratan. Its most remarkable attribute is its size: its wingspan was ca. 1.9-3.8 m, roughly the size of the largest flying birds today (e.g., wandering albatross), and it was not fully grown at death. Triassic and Jurassic pterosaurs have long been stereotyped as relatively small animals, constrained to wingspans of approximately 1.6-1.8 m or less over the first ca. 70 million years of their evolutionary history, 5 before becoming larger in the latest Jurassic or Early Cretaceous, culminating in airplane-sized giants like Quetzalcoatlus with 10-m wingspans. 5 A few tantalizing specimens have hinted at larger Jurassic pterosaurs, [36][37][38] but these are often limited to one or a few bones, which make body size estimations difficult. Dearc is the first Jurassic pterosaur whose wingspan can confidently be estimated at ca. 2.5 m or greater, based on a well-preserved, articulated skeleton. 5 Its closest relatives, Angustinaripterus and Sericipterus, are also sizeable for Jurassic species, with wingspans previously estimated at 1.6 1 -1.7 38 m extrapolated from patchy fossils. Our regression equations indicate larger wingspans for these taxa: ca. 2-3 m, which is still approximately 10% smaller than Dearc. Dearc, therefore, anchors a clade of large, long-snouted Jurassic non-monofenestratans: Angustinaripterini. 38 The large size of Dearc prompted us to re-examine fragmentary specimens from the Taynton Limestone, 39 an English unit that formed at the same general time as the Scottish Middle Jurassic deposits, in or along the margins of the same seaway. We identified 17 specimens-all single bones-that yield wingspan estimates of over 1.7 m based on our predictor formulas (above), including several that may have had wingspans of over 3.0 m. The discovery of Dearc, and our survey of Taynton specimens, reveals that Jurassic pterosaurs were capable of achieving considerably larger sizes than previously thought. Jurassic pterosaurs may still have been constrained in sizeand certainly there is no evidence they approached the grandeur of giant Cretaceous pterodactyloids-but if so, that constraint was at a substantially greater wingspan than 1.6-1.8 m. 5 Trends in pterosaur size evolution, particularly the shift to increasingly larger species in the Cretaceous, have been interpreted in terms of two main hypotheses, which are not mutually exclusive: (1) advances in the pterodactyloid body plan allowed them to become larger and more efficient fliers than non-pterodactyloids, and (2) the diversification of birds (Avialae) may have driven latest Jurassic/Cretaceous pterosaurs into ever-larger size niches. 5 Our identification of Dearc demonstrates that non-pterodactyloids were able to grow to larger sizes by the Middle Jurassic, with some evidence for large pterosaurs back to the Early Jurassic, 37 tens of millions of years before birds underwent their adaptive Cretaceous radiation. 40,41 These size increases seemingly occurred too early for avialans, which are first definitively known from the Late Jurassic, to have been a direct cause. Alternatively, if there was pressure on pterosaurs to become larger, it may have started deep in the Jurassic and involved competition with unrecognized early avialans or other animals, 40 like non-avialan feathered dinosaurs or other pterosaurs.
Not only is Dearc large, but it and its closest angustinaripterin relatives possess derived characters considered keystones of the pterodactyloid skull ''module,'' notably an elongate skull and inclined quadrate. Previous work has argued that the transition between non-pterodactyloids and pterodactyloids involved a nearly perfect modular shift, in which features of the skull changed together in a tightly integrated unit, followed by the body, limbs, and tail. 6 While we do not dispute the overarching modularity of the pterosaur skeleton, Dearc hints that some of the trademark ''pterodactyloid'' features convergently evolved in other groups, 42 perhaps due to feeding ecology or other factors. Yet the endocranial anatomy of Dearc is distinctly primitive, as it has the unflexed brain and smaller flocculus of basal pterosaurs 7,24 and not the transformed brain of pterodactyloids. Thus, it seems, non-pterodactyloids had a similar neuroanatomy regardless of body size.
Our recognition of large Middle Jurassic pterosaurs exposes a taphonomic bias, to add to the already notoriously problematic record of pterosaurs. 43,44 There were fairly large pterosaurs in the Jurassic (wingspans > 1.8 m), but reasonably complete adult or near-mature skeletons are mostly lacking, for reasons unclear and worthy of further study. The Middle Jurassic age of Dearc adds to increasing evidence that this interval-once a frustrating gap in the pterosaur record-was in fact a dynamic time of diversification, in which a variety of basal taxa and early monofenestratan lineages 6,39 coexisted and occupied a range of environments, from open marine to lagoonal, nearshore to desert, around the world. We can now add larger taxa, with derived pterodactyloid-type skull characters, to that roster. As with dinosaurs 45 and mammals, 46 the Middle Jurassic was likely a vibrant time in pterosaur history, not a static and archaic prelude to a Cretaceous explosion of larger, 5 more disparate, 47 more efficient fliers. 20 With the dawn of the Cretaceous, however, the diverse non-monofenestratans disappeared, including larger ones with pterodactyloid convergences. This mystery remains to be solved.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

INCLUSION AND DIVERSITY
The author list of this paper includes contributors from the location where the research was conducted who participated in the data collection, design, analysis, and/or interpretation of the work.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
The main subject of this study, the holotype specimen NMS G.2021.6.1-4, is a well preserved, articulated, and reasonably complete skeleton preserved in a lagoonal bioclastic limestone (Figures 1, 2, 3, S2, and S3). The exposed anterior skull of the specimen was discovered at Brothers' Point, Isle of Skye, Scotland, by AP during a National Geographic-funded fieldtrip led by SLB on May 23, 2017 ( Figure S1). The entire specimen was recovered by the team (including SLB, TJC, AP, DAR, and NDLC) on May 24-25, 2017. The brunt of collection work was conducted by DAR using a diamond-tipped rock saw. All fieldwork was conducted under a permit from Scottish Natural Heritage. The skeleton was found in the Upper Lonfearn Member of the Lealt Shale Formation (Great Estuarine Group), at a site that is referred to as Brothers' Point (BP) 3 in the literature. 11 BP3 is located immediately west of the mouth of Lonfearn Burn into Port Earlish (57.5863 N, 6.1494 W).
Comparative data on other specimens/taxa was retrieved from personal observations and the literature.

Fossil genuineness
As our team discovered the specimen in situ, we can confirm its genuineness. The specimen was mechanically prepared by Nigel Larkin, which involved gluing broken bones together and removing limestone matrix to expose the bones, but no other modifications. During collection and preparation, the fossil was separated into four slabs. The largest slab-what we refer to in the text as the 'main slab'-contains the greatest number of bones (trunk, forelimbs, sacrum and tail bones) exposed in dorsal view. Report slab-what we refer to in the text as the 'main counterslab'-includes partial trunk, tail, pelvic and hindlimb, and left forelimb elements exposed in ventral view (the underside of the slab also contains the second wing phalanx). The third slab contains the third wing phalanx with an articulation point. The fourth slab includes the skull and anterior cervical vertebrae, exposed in dorsal view. The first and second slabs were separated from each other by hammer and chisel during the removal of the skeleton from the field site. The third slab was also separated from the remainder of the skeleton during field excavation. The fourth slab was removed from the first slab by Nigel Larkin during preparation, so that it could be mCT scanned.
Fossil examination NMS G.2021.6.1-4 was studied first-hand in the University of Edinburgh Deep Time palaeontology laboratory by many of the authors, principally NJ and SLB. The specimen was photographed by GFF and NJ using a Nikon D850 camera with Nikkor 14-24 mm and MicroNikkor 60 mm lenses. Where adjustments were made to contrast, brightness, or white balance, they were applied to the whole image. After image adjustment, the background was digitally removed. Some images were created using the automated focusstacking mode of the Nikon D850 to create enhanced-focus images in HeliconFocus v 7.5.5.

Computed tomography scanning
The skull and associated proximal cervical vertebrae were separated from the main slab by Nigel Larkin during preparation. The skull piece was then scanned using a custom-built X-ray mCT instrument at the School of GeoSciences, University of Edinburgh. Multiple, overlapping scans of the skull were acquired at 110 keV,and 50 W target power loading with a 0.9 mm copper and 1 mm aluminum energy filter, using 2 3 2 s exposure per projection (two frames averaged) and 2000 projections per scan. The scans were reconstructed by filtered back projection using Octopus v8.9 software. 48 The geometric voxel resolution of the scans was 97mm. The overlapping scans were registered and merged into a single volume using Avizo (version 9.1) and segmented manually in Mimics (Materialize N.V. 2014). Attempts were made for machine learning segmentation using the Trainable Weka Segmentation extension on Fiji/ ImageJ, but with low contrast between bone and matrix and the presence of pyrite, manual segmentation in Mimics was required. While the mCT scan offered good contrast in regions where the ratio of matrix to bone is low, contrast was reduced in the posterior part of the skull (jugal, postorbital, squamosal), and manual segmentation of this region was challenging. These segmented regions are more interpretive and low-detailed ( Figure 2). Although the posterior skull has reduced contrast and resolution, the internal spaces of the brain and ear region are well defined, making it possible to produce 3D endocasts. The dorsal section of the endocast is missing due to recent beach erosion, and the terminal region anteriorly is difficult to assess as the endocast opens to the vacuity around the orbits. The semi-circular canals can be segmented in part but are of too poor resolution to segment completely.

Bone histology
Thin sections ( Figure S3) were prepared using a modified version of the method outlined by Padian and Lamm. 13 A portion of the wing phalanx was removed at a natural break, and was embedded in Buelher Epothin II epoxy resin under a vacuum (À1 atm). The epoxy was left to cure at room temperature for 48 h and was then cut at the target plane using a Buehler Isomet 1000 Precision Saw and a diamond-tipped wafering blade. The billet and an acrylic slide were each frosted on a glass plate using 600-grit silicon carbide abrasive slurry (Kemet International) to remove saw marks and increase surface area for adhesion. The billet was mounted to the slide using cyanoacrylate adhesive (Loctite) and was left to cure at room temperature for 3 h. The mounted billet was resectioned to 700 mm in thickness using the Buehler Isomet 1000 Precision Saw. The section was then polished by hand on a glass plate using 600-grit silicon carbide abrasive slurry (Kemet International) until optimal contrast was achieved. The slide was finished by polishing using a 1200-grit silicon carbide abrasive slurry (Kemet International) on a glass plate, and then two 10,000-grit lapping pads (Kemet International), the second of which was impregnated with mineral oil (Johnson & Johnson) to improve optical contrast. The slide was observed and photographed using a Leica DMLP Transmitted Light Polarizing Microscope and Leica Application Suite 4. Photomontages were created using the automated 'photomerge' tool in Photoshop Creative Cloud v22.4.1. Osteocyte lacunar density was calculated using the method of Cullen et al. 49

Wingspan estimates
In order to estimate the wingspan of NMS G.2021.6.1-4 and isolated Taynton Limestone specimens, we utilized regression equations that predict wingspan based on the dimensions of single bones. These equations were constructed using data from two genera of basal pterosaurs (non-monofenestratan species closely related to Dearc) known from large samples that span a range of body sizes and ontogenetic stages: Rhamphorhynchus and Dorygnathus. We compiled measurement data for specimens of these genera for which both total wingspan and the lengths of the individual bones in question can be measured. Then, for each bone, we ran a separate regression equation: one using Rhamphorhynchus data only, the other using Dorygnathus data only. All regressions were ordinary least-squares (OLS) regressions, performed in PAST, on log-transformed data. The source data for each regression, the resulting regression equations, regression plots (with 95% confidence intervals), and wingspan estimates for Dearc and isolated Taynton bones are provided in Data S1C-S1Q. Further explanation is given here:

Postcranial Bones
In order to estimate the wingspan of NMS G.2021.6.1-4 and Taynton Limestone pterosaurs, we carried out regression analyses using datasets consisting of the lengths of complete individual elements (humeri, ulnae, wing phalanges, and femora) and their corresponding complete wingspans. The humeral data were previously published in Appendix 1 of O'Sullivan et al. 37 Each one of the Jurassic non-monofenestratan pterosaur specimens selected has at least one complete wing, which allows us to accurately measure the absolute wingspan. (Wingspan is defined as the length of one wing doubled. The body is not included as pterosaurs flew with the humerus and radius/ulna complex held in a v shape, the width of which is approximately equal to the lateral width of the body). We performed a total of 12 postcranial regressions: humerus, ulna, wing phalanx 1, wing phalanx 2, wing phalanx 4, and femur, in each case for Rhamphorhynchus and Dorygnathus separately. In each regression, we plotted the logarithms of wingspans along the y axis against the logarithms of lengths of a complete postcranial element (e.g., humerus, wing phalanx 1, etc.) on the x axis. For each regression, we then produced a line of best fit using OLS (with 95% CIs) and measured the variance of the data using the R 2 score and the significance of the linear relationship with a permutation p value. We then took the Linear Regression Equation from our plots and used it to estimate wingspan using the formula below.
We utilized the same regression methodology to use the skull length to estimate wingspan, but using only Rhamphorhynchus, as too few complete skulls of Dorygnathus are available to conduct a robust regression analysis.
Finally, we used our postcranial and cranial regressions to estimate the wingspan of two close relatives of Dearc: Sericipterus and Angustinaripterus. They have been described as large Jurassic pterosaurs, but their wingspans were estimated previously at 1.6-1.7 m based on fragmentary fossils and without explicit regression equations.

Stratigraphy, dating, and taphonomy
The Lealt Shale Formation ( Figure S1) is Middle Jurassic in age. It is part of a package of formations known collectively as the Great Estuarine Group, which were formed during repeated cycles of delta progradation and retrogradation into marginally marine lagoonal systems between ca. 166-170 million years ago. 9 The dating of these rocks has historically been challenging because they do not preserve ammonites-the most relevant index fossils-in abundance. 9,50 However, their stratigraphic relationships with the underlying Bajocian Garantiana Clay Member and Bearreraig Sandstone Formation and the overlying Callovian Staffin Bay Formation constrain the age range for the Great Estuarine Group-and hence the Lealt Shale-to latest Bajocian to Bathonian. [50][51][52][53] The dating of the Lealt Shale in particular has been considered Bathonian, based on correlation with neighboring formations and the presence of ostracods in the overlying Duntulm Formation, which bear similarities to those of the progracilis zone. 9,10 If accurate, this would mean the Lealt Shale is coeval with the pterosaur-rich Taynton Limestone of England.
During the Bajocian-Bathonian, Britain was covered by a shallow epicontinental sea and divided into three major landmasses: London-Ardennes, Cornish-Mendip, and the northern Scottish-Pennine; the latter of which was where the Lealt Shale and other Skye deposits formed. 51 Around the end of the Bajocian, the offshore Hebridean Basin experienced sea-level rise, and became linked to neighboring basins by flooding of low grounds, which formed shallow areas of deposition with fluctuating salinity levels. 52 These events are represented by the lagoonal/marginal marine Lealt Shale and the deltaic Valtos formations of the Great Estuarine Group, both of which are well exposed on Skye. Much of the Hebridean Basin was very shallow at this time, and submerged areas closer to shore underwent periodic drying and subaerial exposure (as evidenced by terrestrial units with dessication cracks and salt deposition) 11 . The local climate during this time was warm and humid. 54 The Lealt Shale (referred to as Estheria or Mytilus Shale in earlier literature), 10 is defined by an abrupt lithographic change from coarse-to-pebbled sandstone at the base (underlying Elgol Sandstone Formation) to bituminous, silty shale. The upper boundary is more gradational, turning from fissile muds to silts (overlying Valtos Sandstone Formation). There are notable changes in index fossils, exemplified by the disappearance of Cyzicus clam shrimp, and appearance of Neomiodon bivalve shells. 51,52 The Lealt Shale is tentatively divided into two members defined by the presence and abundance of Mylitus shells. The base is formed of monotypic Praemytilus strathairdensi shell-beds. This unit (the Kildonnan Member) is best exposed on the Isle of Eigg, as a gray silt shale, with occasional beds of limestone, and at least two bone beds and a coarse sandstone interval. The second member-the Lonfearn-is predominately composed of shales with ostracods, gastropods and the index fossil Cyzicus. 10 The holotype of the new pterosaur Dearc comes from the Lonfearn member.
The Lonfearn member is composed of three distinctive facies. 9 The fissile mudstone facies have shell plasters, high (4.5% max.) total organic content 55 and lack bioturbation suggesting low water energy and low oxygen levels. The shell limestone facies is composed of monotaxa Neomiodon shell beds replaced with biosparite and displaying effects of winnowing, 56 with variable levels of compaction. The new pterosaur skeleton originates from the ferruginous oolitic limestone facies, predominately composed of ooids with ferroan dolomite coating, with local muddy interbedding, pervasive fibrous calcite veins, disarticulated and winnowed bivalve shells and ostracods valves, which are found within a layer below the pterosaur specimen.
The pterosaur comes from a bed infilling sauropod trackway on a laminated micrite. It is a well-sorted, bioclastic limestone. 11 The bioclastic elements are chiefly disarticulated Neomiodon shells with some ostracods and conchostracans, although none have been found in the immediate vicinity of the skeleton. There are desiccation cracks present on the bed, set in some distance from the fossil, suggesting localized, periodic drying up periods in a shallow-water setting. 9 The skeleton is remarkably well preserved, with delicate and hollow bones mostly uncrushed and preserved in three dimensions, and it shows no evidence of substantial fracture, compaction, or disarticulation, suggesting it was buried in a low-energy environment, which dePolo et al. 11 suggested as biologically active mudflats.

Report
The rock matrix within which the pterosaur skeleton resides is rich in evenly distributed diagenetic cubic pyrite (up to 3 mm), which surrounds and in some cases (e.g., left metacarpal) extends into the fossil. The Lealt Shale Formation is intruded by numerous dolerite sills but the pterosaur skeleton has not been visibly altered by diagenesis related to magmatism and metamorphism.

Phylogenetic analysis
The phylogenetic relationships of Dearc sgiathanach among pterosaurs were tested using an analysis of discrete anatomical characters in a parsimony framework. The analysis includes 53 in-group terminal taxa and five outgroup taxa, scored for 155 unordered and equally weighted characters (Data S1B); we do not order multistate characters, to avoid any assumptions of transformation sequence. Characters include those assembled from numerous published pterosaur phylogenies and novel characters that particularly relate to features that are well preserved in Dearc (e.g., palatal, endocast, and ceratobranchial characters 71-79). Assembled characters originate from first-authored studies of Andres; 25 Wang 22 [details in Data S2]. Characters exhibiting strong ontogenetic variation were excluded. Our taxon sample was constructed to focus on basal non-monofenestratan pterosaurs (the grade into which Dearc falls) and to mostly exclude taxa known only from immature specimens, which along with ontogenetically highly variable characters can confound phylogenetic analysis. We also excluded some taxa known only from highly incomplete or very poorly preserved specimens. We do note that we have included some taxa that are known mostly, or exclusively, from non-adult material or have immature type specimens: most notably Dearc itself, but also Triassic taxa like Eudimorphodon and kin, and Anurognathus. We elected to include such taxa because of their potential phylogenetic importance (they anchor major clades or fill important gaps in the stratigraphic record) and because specimens, although immature, exhibit good preservation and are relatively complete.
Fundamentally, we acknowledge that our phylogenetic analysis relies on a series of methodological choices, and other authors may prefer alternatives. Our main goals were to focus on the relationships of non-pterodactyloids and to minimize false phylogenetic signals that could derive from immaturity and incompleteness of specimens-a long-standing problem with pterosaurs. One alternative to the latter problem is to score ontogenetic-dependent characters as missing data ('?') for species with unknown mature specimens. However, we prefer not to use this approach, because it can be challenging to recognize ontogenetic stages in pterosaurs due to the plasticity of various indicators of maturity (bone fusion and ossification, body size, histology). For instance, without our histological evidence, we would probably have considered the Dearc a mature (or near-mature) adult, not a still-growing juvenile or sub-adult. Given such uncertainty, we prefer the more straightforward approach of simply removing specimens/taxa known from what are clearly very immature specimens or which are highly incomplete.
We performed a maximum parsimony analysis in TNT. We began with a New Technology Search (RSS, CSS, Ratchet, Drift, Tree Fusion with default settings), in which the minimal length tree was identified 50 times. A most parsimonious tree length of 491 was found 362 times. These trees were then subjected to an additional round of TBR branch swapping, which found a total of 9000 most parsimonious trees (overflow) most parsimonious trees (consistency index = 0.440, retention index = 0.731). We used these trees to conduct a strict consensus in which 27 taxa were excluded a posteriori, in order to eliminate wildcard taxa and specimens with a suspected preponderance of juvenile morphological scores. Clade support was determined using Bremer and bootstrap values. The strict consensus recovered Dearc in a clade with Angustinaripterus longicephalus and Sericipterus wucaiwanensis; this clade is sister to Rhamphorhynchus muenstri.