Cretaceous amniote integuments recorded through a taphonomic process unique to resins

Fossil records of vertebrate integuments are relatively common in both rocks, as compressions, and amber, as inclusions. The integument remains, mainly the Mesozoic ones, are of great interest due to the panoply of palaeobiological information they can provide. We describe two Spanish Cretaceous amber pieces that are of taphonomic importance, one bearing avian dinosaur feather remains and the other, mammalian hair. The preserved feather remains originated from an avian dinosaur resting in contact with a stalactite-shaped resin emission for the time it took for the fresh resin to harden. The second piece shows three hair strands recorded on a surface of desiccation, with the characteristic scale pattern exceptionally well preserved and the strands aligned together, which can be considered the record of a tuft. These assemblages were recorded through a rare biostratinomic process we call “pull off vestiture” that is different from the typical resin entrapment and embedding of organisms and biological remains, and unique to resins. The peculiarity of this process is supported by actualistic observations using sticky traps in Madagascar. Lastly, we reinterpret some exceptional records from the literature in the light of that process, thus bringing new insight to the taphonomic and palaeoecological understanding of the circumstances of their origins.


Scientific Reports
| (2020) 10:19840 | https://doi.org/10.1038/s41598-020-76830-8 www.nature.com/scientificreports/ patterns can provide information for the determination of mammalian taxa 18,22,23 . The scale pattern can differ between mammalian species, between individual strands of the same specimen, between stages of ontogenesis and along the length of a single strand depending on relative proximity to the basal region or the tip 21,24 . Furthermore, it is also possible for different species to show similar patterns 21,24 . The cuticular scale pattern is genetically controlled 23,25 . Its morphology and arrangement depend on biomechanical forces, which act on the cells during their hardening through the keratinisation process, and on growing speed 26 . Until now, the oldest described mammalian hair in amber corresponds to two strands from the late Albian French amber of Archingeay-Les Nouillers 16 . Mammalian hair is also reported from the early Cenomanian Burmese amber (northern Myanmar), even retaining the scale pattern 27 . One specimen of filamentous structure from the Santonian amber of Yantardakh (Taimyr, Russia) could also be a mammalian hair strand 15 . The youngest Cretaceous mammalian hair comes from the late Campanian amber of Grassy Lake in Canada 6 , but it is still undescribed. The record of mammalian hair in Cenozoic amber is comparatively richer (e.g., 18,28,29 ). Based on its molecular composition, the preservation potential of feathers and hair is high 30 , although the keratin protein has a low fossilisation potential 10,31 . Dinosaur feathers from compression sites are well-known 30 , although the three-dimensional preservation in amber allows an exceptional visualisation of their anatomic structure 7 . Mammalian hair in Mesozoic mammaliaforms found in compression sites has been preserved as a halo of carbonised fur or tufts (e.g., 19,32 ), but usually the scale pattern is obscure or poorly preserved (e.g., 33 ). Furthermore, the taphonomic conditions that facilitated the preservation of vestitures in amber are little-known, and they have been usually related to resin flows produced near the forest floor (or 'litter amber' sensu 34 ), requiring at least a contact of the individual before escaping 7 . In contrast, the taphonomic processes of insect preservation in amber have been widely studied 35,36 . Actualistic methods as sticky traps have been previously used to evaluate the accuracy of the record of forest arthropods in resin and amber 3,37 , but this kind of traps also records vertebrate vestitures and can provide an example of how resin traps these remains.
Here, we report on two Cretaceous amber pieces from Spain, containing a grouping of avian dinosaur feather remains and three strands of mammalian hair, respectively, resulting from the recording of vertebrate vestiture through a distinct biostratinomic process that we describe for the first time and that is unique to resins.

Results
Avian dinosaur feather remains. A small flake (4.1 mm long and 2.8 mm at its greatest width) of an aerial, stalactite-shaped amber piece (CPT-4200, from San Just outcrop, Teruel, Spain) contains abundant feather remains (the barbs are easily recognised) of an avian dinosaur (Fig. 1). The flake piece is curved in cross section, with the convex (external) surface being dark and the concave one being of the same colour and transparency as the rest of the piece (Fig. 1a). These features indicate that it is a fragment of one of the concentric layers, originating from a resin flow; both the internal core of the piece and a potential later layer (or layers) were lost. The external surface is darker due to original desiccation under aerial conditions. This surface is completely covered by feather barbs dispersed at random without preferential orientation. The barbs (at least a few millimetres long and ca. 0.29 mm wide, with barbules ca. 0.26 mm long; Fig. 1a-d) are not impressions in the surface, but conserved original organic matter, although some small patches are missing (Fig. 1c). Barbs and barbules are well preserved, including the nodes and internodes of the pennula (Fig. 1e,f). The sticky external resin layer must have already been somewhat hardened when the feathers came into contact with it because it does not show deformation from contact and the barbules did not penetrate the resin but are present over the whole original convex surface, as preserved.
Mammalian hair strands. Three mostly parallel-aligned strands of mammalian hair are present in the fragment AR-1-A-2019.88.1 (from Ariño outcrop, Teruel, Spain) (Figs. 2a, S1) (see the description of the whole amber piece AR-1-A-2019.88 in the Supplementary information). They are named X, Y and Z and are exceptionally well preserved, even showing the surface scale pattern (Fig. 2b-d). The scale pattern is easily observed due to the piece having been broken during its extraction, which has facilitated its drawing and SEM imaging (Fig. 3), in contrast to the two hair strands in the French amber, which are poorly preserved 16 , and the hair records found in compression sites (e.g., 19,33 ). The three hair strands are brown and straight, and apparently they were rigid. Strand Y is broken, and it is possible that the two fragments are actually from two different strands. There is a discontinuous gap between them where the amber piece seems to be broken, but as they are in the same plane below strand X, we treat them as two fragments of the same strand. The strands are incomplete, and the tips and hair follicles are not preserved, so the orientation of the scale arrangement is not possible to determine. The scale pattern can be partly seen in the external brown surface, but the transparency of the amber fragment facilitates visualisation of the scale pattern mainly as impressions (or cast prints) in the amber, as shown on the internal surfaces of the broken sections. Strand X is 6.72 mm long and around 0.07 mm in diameter; broken strand Y is composed of two strand fragments which are 3.26 mm and 3.20 mm long, respectively, with a diameter of around 0.07 mm, and strand Z is 6.19 mm long and around 0.07 mm in diameter. The cross section of the hair strands cannot be visualised, so medulla and cortex cannot be seen, and are possibly not preserved along the strands, although they could be in those parts where the brown surface of the cuticle is present. An accurate description of the scale pattern was achieved based on the clearly marked scale margins appearing as impressions in the amber.
The scale pattern and proportions of the three hair strands are similar. Scales are wider than they are long and are in a transverse position in relation to the hair longitudinal axis (Figs. 2b-d, 3). Scale length varies between 5-7 µm. The surface scale pattern is wavy in all three strands, mostly regular in strand Y and slightly irregular in strands X and Z. Scale margins are smooth and close together. Following the terminology of Chernova 23   www.nature.com/scientificreports/ C. madagascariense, respectively. The strands were pulled off retaining the proximal and distal portions. Sticky traps located at 0 m high in the trees recorded hair strands and tufts of rodents that are most likely non-arboreal (Fig. 5e); several of these sticky traps were even gnawed (Fig. 5c,d). Hair, most likely from lemurs, is found in sticky traps at 2 m high in C. madagascariense (Figs. 4h, 5g). In contrast, feathers are not found adhered to the traps.

Discussion
Most isolated feathers found in Burmese amber have been determined as belonging to Enantiornithes based on the presence of skeletal remains with associated feathers of this group (e.g., 38), although this hypothesis is controversial due to the many feather morphotypes that are not associated with skeletal remains 7 . The San Just feather remains could be consistent with this taxonomic affinity, but the absence of enantiornithine skeletal records in Spanish amber precludes even a tentative determination. Filamentous elongated remains are often found in amber and they can be confused with hair strands. Fungal mycelia have been reported in ambers from different deposits, showing hyphae of 2-8 μm diameter (e.g., 39,40 ), as opposed to the Ariño strands which are around 70 μm in diameter. Undetermined plant fibres also appear in amber, but they are usually distinct from hair-like structures, as they can be very irregular in shape. The key characteristic of hair strands is their scale pattern 22,23 , although Chernova 14 noticed a similarity between the microstructures of determined hairs and feathers as a result of morphological convergence. The Ariño strands clearly correspond to mammalian hair as their scale pattern is very similar to others from fossil (Mesozoic and Cenozoic) and extant hair specimens (Figs. 5e-g, 6).   33 . Hair strands of S. xenarthrosus include two morphotypes of scale patterns, as the primary hairs show imbricate, ovate scales forming an irregular mosaic and the secondary hairs possess annular scales embracing the shaft with simple or serrated free margins (extended data Fig. 5 in 33 ). The scale pattern of the Ariño hair strands does not correspond to the morphotypes of the Las Hoyas specimen, as more than one scale embraces the shaft and the pattern is wavy with smooth margins. Considering extant taxa, the scale pattern of mustelids (Carnivora: Mustelidae) is very similar to that of the Ariño hair strands, for example, Martes martes Linnaeus, 1758 (plate 72-80 in 22 ) and the wavy pattern at the pars basalis of Lutra lutra Linnaeus, 1958 (Fig. 7 in 42 ). Despite all the above knowledge, accurate determination of the hair strands from Ariño is not also possible, due to the poorly preserved scale pattern of fossil hair in Cretaceous mammalian specimens. The Ariño hair strands from the early Albian are the oldest mammalian record in amber and increase the previous known Cretaceous record of mammalian hair confined to late Albian-early Cenomanian ambers 16,27 . New descriptions of scale patterns of hair in Cretaceous amber, whose preservation capacity is exceptionally high, as seen in the Ariño hair strands, could shed further light on the evolution of this mammalian character through time.
Based on the matching scale pattern and proportions of the three hair strands, their proximity and alignment in the amber piece, we assume that they were a tuft. Although they are only three strands, their disposition and the lateral similarity in diameter and scale pattern support our inference that they are not randomly detached www.nature.com/scientificreports/ strands. This is a unique record in the Cretaceous, as other hair strands in amber (France and Myanmar) appear as isolated specimens 16,27 . The two hair strands in French amber are in the same piece, separated by only 1 mm, but they cannot be confirmed as a tuft because of their poor preservation 16 . Therefore, the Ariño hair strands show more similarities with the record of tufts in the Eocene Baltic amber (e.g., 28,29 ) (Fig. 6). Both amber pieces presented in this study correspond to resin flows exposed to aerial conditions for a time, based on their morphology and the presence of surfaces of desiccation. Despite the small size of the amber piece CPT-4200 (showing feather remains), the manner in which the barbs cover the whole surface of desiccation, as preserved, indicates that this record was not the consequence of contact between a detached feather, or feathers, and the sticky resin emission. This record can best be explained thus: an avian dinosaur made contact with the www.nature.com/scientificreports/ resin emission, resting there for the necessary amount of time, for example during sleep, for the more external barbs of its vestiture to become firmly fixed in the hardened resin. Later, with the movement of the animal, the fixed barbs detached. However, it can be observed that some portions of the barbs were apparently better fixed than others as there is differential preservation along the barbs (Fig. 1c). Regarding the amber fragment AR-1-A-2019.88.1 (showing the hair tuft), the hair strands could have become embedded slowly while the mammal in question was resting or sleeping near the resin source, in similar fashion to the feathers in piece CPT-4200. The two amber records described herein are very peculiar and their origins are far from the common process for generation of bioinclusions in resins that are preserved as amber or copal. Typically, a bioinclusion originates when an animal or a detached organismal remnant makes contact with fresh, sticky resin and is thus trapped and then embedded. Here, we present the "pull off vestiture" biostratinomic process unique to resins, which we define as follows: "the entrapment of external portions of vertebrate vestiture, more specifically small portions of plumage and pelage, of living individuals that had rested for a time in contact with a fresh, sticky resin emission that hardened and retained those vestiture portions". Note that this process does not imply the death of the animal. The process requires a type of resin that can harden very quickly after being exuded. Such a feature has been observed in extant resins from Agathis australis (D. Don) Loudon, 1829 (gymnosperm) in New Zealand  (Figs. 4, 5). These records are similar to those of the natural "pull off vestiture" process of resins, although the trapping capacity of the sticky traps is much higher. Hair would be trapped immediately in the sticky traps with minimal contact, whereas the "pull off vestiture" process involving resin requires contact to be sustained for the time it takes for the resin to harden. Squamate reptile scaly skin was also recorded on the sticky traps (Fig. 4i), as has been found in Cretaceous amber from France, Lebanon and Myanmar [43][44][45][46] , but these kinds of record would be recorded differently as they could be explained by detached portions of lizard moults which do not require high stickiness. The "pull off vestiture" biostratinomic process described herein is unique to resins and clarifies some intriguing findings in Cretaceous and Cenozoic ambers. Other semi-fluid preservative materials, like asphalt (highly viscous liquid or a semi-solid form of petroleum) or tar seeps, do not pull off vertebrate vestiture while allowing the survival of the individual 47 . Paired ornamental feathers of enantiornithine tails in the manner of their distal parts in relative live position (each pair) from the Burmese amber record have been reported 48,49 . Xing et al. 48 commented that the paired feathers, which they found abundantly, could easily be removed, maybe as a sacrificial gesture in defensive behaviour. The record of these abundant findings could be related to the "pull off vestiture" process, as are other enantiornithine feather records in amber (e.g., 7 ). The tail feathers could become stuck in the resin while the individual was resting or sleeping close to the resin source in the tree, and after a time, when the resin was hardened and the individual left, the feathers were removed, similar to the biostratinomic process of the San Just and Ariño records in this work (Fig. 7). The tufts found in Eocene Baltic amber only preserve the distal parts of the strands 29 and are coincident with the "pull off vestiture" process (Fig. 7). The basal parts of the hairs (including the follicles) are not preserved in the amber pieces because typically only the distal parts contacted and became embedded in the resin emissions to undergo the "pull off vestiture" process. Records of mammalian hair strands in Miocene Dominican amber correspond to isolated strands with different orientations, www.nature.com/scientificreports/ not forming tufts, and even retaining basal parts (e.g., 18,50 ), so they could have been blown by the wind or shed randomly by the mammalian individual and do not correspond to a "pull off vestiture" process. The rich record of mammalian tufts in Baltic amber could be related to the presence of arboreal mammals such as sciuromorph rodents 51 , although Sidorchuk et al. 29 proposed the floor-dwellers, erinaceomorph amphilemurids. The extreme scarcity of hair in Cretaceous amber suggests that the arboreal mammals did not inhabit the resiniferous forests during this period, although an arboreal lifestyle has been suggested for gliding Mesozoic mammals from compression sites (e.g., 32,52 ). The new biostratinomic approach described herein can explain records of vertebrates in amber and provide new inferences of palaeoecological information about the ancient resiniferous forests.

Material and methods
The San Just outcrop is located near Utrillas (Teruel, Spain) and corresponds geologically to the Middle or Upper Member of the Escucha Formation 53,54 . It has been dated as middle-earliest late Albian based on palynological data, and related to a swamp plain environment 55 . It is the type-locality of 23 species and its diverse record of bioinclusions contains coprolites, fungi, plants, arachnids, 12 insect orders and dinosaur feathers 53,54 . More data about the amber piece CPT-4200 are available in the Supplementary information. The Ariño amber outcrop is in the Santa María opencast mine near Ariño (Teruel, Spain), which has provided a diverse fossil record from the bonebed layer AR-1, including coprolites, algae, plants, molluscs, ostracods, fish, turtles, crocodiles and dinosaurs 56 , and is the type-locality of nine new taxa to date. This layer, belonging to the Middle Member of the Escucha Formation, has been dated as early Albian based on the charophyte assemblage 57 and related to a freshwater swamp plain in a deltaic-estuarine system with salinity variations due to marine inputs under subtropical-tropical climate conditions 58 . The amber piece AR-1-A-2019.88 with the mammalian hair studied herein was found in AR-1, composed of marls within an alternation of marl and limestone levels with alkaline oligotrophic origin, pond or shallow lake, and was unearthed from the layer in separate fragments; a description of the piece is available in the Supplementary information (Fig. S1).
The San Just piece CPT-4200 was photographed with a digital camera Olympus Camedia MODEL N.C5050 ZOOM attached to an Olympus SZX9 stereomicroscope, and the detailed micrographs of the barbules, using a ColorView IIIu Soft Imaging System attached to an Olympus BX51 compound microscope; the frontal view of the feather remains (Fig. 1a) was taken using a digital camera Canon EOS 650D and z-stacked automatically by the software Macrofotografía version 1.1.0.5 (www.macro rail.com). A compound microscope Olympus CX41 equipped with an attached camera lucida tube and a digital camera sCMEX-20 was used to make the drawings and the micrographs of the mammalian hair strands and other syninclusions in piece AR-1-A-2019.88; these micrographs were processed with ImageFocusAlpha version 1.3.7.12967.20180920 (www.eurom ex.com). SEM images of the fossil and extant hair strands with gold sputtering were obtained with a Hitachi S4800 Electronic Microscope at the Microscopy Services (SCSIE) of the University of Valencia (Spain). Figures were prepared with Photoshop CS2 version 9.0 (www.adobe .com) and Photoshop CS6 version 13.0 (www.adobe .com). The description of the scale pattern of the hair strands follows the nomenclature of Teerink 22 and Chernova 23 .
Regarding the Baltic amber pieces (Fig. 6), they are housed at the Senckenberg Research Institute and Natural History Museum (SMF) in Frankfurt (Germany). The colour photographs and Z-stacks images were performed under a Nikon SMZ25 microscope, using Nikon SHR Plan Apo 0.5x and SHR Plan Apo 2x objectives with a microscope camera Nikon DS-Ri2 and the NIS-Element software version 4.51.00 (www.micro scope .healt hcare www.nature.com/scientificreports/ .nikon .com). Infrared reflected photomicrographs (black and white) were taken with a Nikon Eclipse ME600D. All these methods were undertaken at the SMF. The actualistic data were obtained in Madagascar with sticky traps following the methodology of Solórzano Kraemer et al. 3 . The sticky traps were located around 12 trees of the species Hymenaea verrucosa (Fabaceae) and four trees of the species Canarium madagascariense (Burseraceae) at 0, 1 and 2 m high during eight days in Sacaramy area (Antsiranana), Ambahy community (Nosy Varika, Mananjary region), Analamandrofo forest, in Andranotsara at 40 km south of Sambava city and Ranomafana National Park, during the 2013, 2015 and 2017 campaigns 59 and were photographed with a Canon EOS 40D digital camera. Hymenaea is a genus of resiniferous tree that originated amber deposits in several places such as Mexico and Dominican Republic. Nowadays, representatives of this genus are found in Madagascar, where an actuotaphonomic work was addressed to study how the resin traps biological remains using sticky traps. The resiniferous genus Canarium occurs in Malagasy forests with dissimilar environmental conditions, thus allowing the comparison of resin trapping in different ecosystems.

Data availability
All data that support this study are available in the main text or in the Supplementary information. Institutions that host the Spanish amber pieces are indicated in the Supplementary information. Actualistic data from sticky traps are available at the SMF in Frankfurt (Germany). www.nature.com/scientificreports/