Preservation and taphonomy of colouration in dinosaurs and their close relatives

Abstract

Over the last two decades, a huge increase in the excavation of exceptionally preserved dinosaur fossils has revolutionised palaeontological research. These fossils retain soft tissues (feathers, scales, skin, and hair) with high micro and nanostructural fidelity. The preservation of the pigment melanin (embedded in tiny vesicles called melanosomes) in these soft tissues permits the reconstruction of in vivo colouration and patterning in long extinct vertebrates. Therefore, this has created unique opportunities to articulate and test hypotheses on life history traits, ecology, behavioural strategies, ancient ecosystems and make predictions supported by strong empirical evidence. However, the scientific arena of ‘palaeocolour reconstruction’ is still in a nascent stage with considerable potential for expansion and refinement. First, although many different methods have been used to estimate palaeocolour but there is no consistent repeatable framework in place which accounts for different preservational biases. Second, statistical models currently used to estimate palaeocolour from melanosome shape data have several shortcomings which need rectification and improvement. Third, despite melanin being the proverbial clef de voûte in palaeocolour studies, much is unknown about the exact taphonomic processes contributing to the exceptional diagenetic stability and soft tissue preservational biases in the fossil record. Lastly, the diagenetic pathways and preservation potentials of non-melanin pigments (like carotenoids and psittacofulvins) in the fossil record has remained unexplored. In this thesis, I focus primarily on these specific research foci. Chapter 1 involves a comprehensive review of pigmentary systems, colour mechanisms in extant and fossil amniotes, and the current ambit of palaeocolour research. This allowed me to develop a new synthetic workflow which collates and contextualises the various methods (e.g., electron microscopic imaging, chemical and statistical methods) currently in use for palaeocolour reconstruction, making the framework robust, highly repeatable, and less liable to be affected by taphonomic bias. In Chapter 2, I compile a large database of melanosome shape in vertebrates rom published work and applied machine learning to improve the accuracy of fossil colour prediction. This new approach has the potential to predict palaeocolour with better accuracy, and to be refined as more data are amassed. Finally, in Chapters 3 and 4, I investigated the effects of diagenesis on melanin, carotenoid and psittacofulvin pigmented feathers of different colours by subjecting them to a special maturation setup which closely mimics natural diagenesis. The results showed that with increased temperature and pressure, melanins start forming inter- and intramolecular crosslinks which allows them to survive and to be preserved as fossils. On the other hand, increased temperature and pressure leads to breaking of alternate double bonds in carotenoids and psittacofulvins but leaves behind a cache of aliphatic, alicyclic, and aromatic products. Additionally, the patterns of diagenesis of carotenoids and psittacofulvins are possible to detect by the combined use of optical/vibrational spectroscopy and mass spectrometry. My work in this thesis, therefore, advances the field of palaeocolour research by contributing to two priority areas, i.e., geochemical preservation of pigments and the development of more accurate, high throughput data analysis methods for inferring fossil colour.