A Model Based on Curvatures of Extant Avian Ungual Bones

Home , Balaur bondoc, Borogovia, Changyuraptor, Dakotaraptor, Eosinopteryx, Pedopenna

Inferring lifestyle for Aves and Theropoda: A model based on curvatures of extant avian ungual bones

A thesis submitted to the University of Manchester for the degree of Master of Science by Research in the Faculty of Science & Engineering

2019

Savannah E. Cobb School of Earth and Environmental Sciences Contents

List of Figures...... 4-5 List of Tables...... 6 List of Abbreviations...... 7-8 Abstract...... 9 Declaration...... 10 Copyright Statement...... 11 Acknowledgements...... 12 1 Literature Review...... 13 1.1 Avians, avialans, and theropod dinosaurs...... 13 1.2 Comparative study and claws...... 18 1.2.1 Claws, locomotion, and ecology...... 19 1.2.2 Quantifying claw form...... 22 1.2.3 Quantifying behaviour...... 27 1.2.4 Comparability across taxa and structures...... 28 1.2.5 Internal structure of the claw...... 31 1.2.6 Techniques for quantitative analysis...... 33 2 Introduction…...... 34 3 Materials and Methods...... 39 3.1 Development of the model...... 39 3.1.1 X-ray techniques...... 41 3.1.2 Geometric measurements...... 45 3.1.3 Statistical analysis...... 48 3.2 Application of the model...... 52 4 Results...... 54 4.1 Measurements and raw data……………………………………….…………..54 4.2 Relationship with body mass...... 63 4.3 Relationship with phylogeny...... 70 4.4 Claw geometry and behavioural category...... 73 4.5 Comparison with fossil taxa...... 84 2

5 Discussion...... 90 5.1 Findings on claw ecomorphology...... 90 5.2 Potential sources of error...... 92 5.2.1 Effects of scaling on form...... 92 5.2.2 Phylogenetic influence on form...... 95 5.2.3 Overlapping ranges: form and behaviour...... 97 5.2.4 Conflicting predictions...... 99 5.2.5 Incomparable fossil morphologies and behaviours...... 102 5.3 Predictions for fossil taxa...... 103 5.3.1 Archaeopteryx...... 105 5.3.2 Maniraptoran dinosaurs...... 106 5.3.3 Avialans and avians...... 112 6 Conclusions...... 115 References...... 119-129 Appendix……………………………………………………………………………….130-141

A Radiographs………………………………………………………………….130

B Photographs of fossil claws………………………………………………….139

Word count (including tables): 26,615

List of Figures

1.1 Maniraptoran taxa with apparent gliding adaptations...... 15 1.2 Proposed mode of climbing for theropod dinosaurs...... 16 1.3 Methods of quantifying claw curvature...... 23 1.4 Methods of calculating claw arc……………………………………...... 24 1.5 Internal structure of the claw...... 27 2.1 Claw radius versus claw arc……………………………………………...…………...37 3.1 The Nomad Pro Radiography Unit.…………………………………………………...42 3.2 Radiographs showing layering process...... 43 3.3 Positions for operating the Nomad Pro Radiography Unit...... 44 3.4 Methods of determining claw angle for sheaths and ungual bones…………………...46 3.5 Alternate method of measuring ventral curvature to the exclusion of the toe pad……47 3.6 Reconstructed fossil ungual bones...... 53 4.1 Histograms showing distribution of body mass and log-transformed body mass for extant avians…………………………………………………………………………..64 4.2 Box plots showing relationship between body mass and ecological group…………...67 4.3 Results of Abouheif’s test for phylogenetic autocorrelation in claw angles……….....71 4.4 Results of Abouheif’s test for phylogenetic autocorrelation in ecological group...... 72 4.5 Curvatures of D-III ungual bones and claw sheaths for all extant taxa……………….76 4.6 Radiographs of avian D-III claws exhibiting significant morphological disparity…...77 4.7 LD axes based on curvatures of D-III ungual bones and claw sheaths for all extant and fossil taxa……………………………………………………………………...………80 5.1 Regression plots for claw angle against body mass...... 93 5.2 Right and left pedal D-III claws of specimens that received conflicting predictions……………………………………………………………………………100 5.3 Specimen of Microraptor gui (IVPP V17972) with remains of enantiornithine bird preserved in abdominal cavity and close-up of line drawing of IVPP V17972……..110 A.1 Accipiters…………………………………………………………………………….130 A.2 Falcons……………………………………………………………………………….130 A.3 Strigiformes………………………………………………………………………….131 A.4 Palaeognathae………………………………………………………………………..131 4

A.5 Otidiformes…………………………………………………………………………..131 A.6 Water birds…………………………………………………………………………...132 A.7 Cuculiformes…………………………………………………………………………132 A.8 Galliformes…………………………………………………………………………..132 A.9 Gruiformes………………………………………………………………………...…133 A.10 Parrots………………………………………………………………………………..133 A.11 Hornbills, rollers, and kingfishers…………………………………………………...133 A.12 Musophagiformes, Caprimulgiformes, Pterocliformes, Opsithocomiformes………..134 A.13 Columbiformes………………………………………………………………………134 A.14 Corvids………………………………………………………………………………134 A.15 Woodcreepers…………………………………………………………………...135-136 A.16 Toucans…………………………………………………………………………136-137 A.17 Woodpeckers and barbets……………………………………………………………137 A.18 Squamates……………………………………………………………………………138 B.1 Avialans……………………………………………………………………………...139 B.2 Dromaeosaurids……………………………………………………………………...140 B.3 Troodontids…………………………………………………………………………..140 B.4 Anchiornithids……………………………………………………………………….141 B.5 Ornithomimidae, Tyrannosauridae…………………………...... 141

List of Tables

4.1 Sampled extant taxa………………………………………………………………..54-59 4.2 Sampled fossil taxa...... 59-60 4.3 Variability between methods of measuring claw angle...... 61-62 4.4 Shapiro-Wilk normality test on claw angles and body mass...... 63 4.5 Linear regressions between body mass and claw angle...... 64-65 4.6 ANOVAs and Tukey tests for body mass and ecological group…………………..68-69 4.7 ANOVAs for body mass and phylogenetic order…………………………………69-70 4.8 Results of Abouheif’s test for phylogenetic autocorrelation in claw angles……...…..72 4.9 Results of Abouheif’s test for phylogenetic autocorrelation in ecological group…….73 4.10 Variation in claw angle data...... 75-76 4.11 Results of non-parametric tests performed on subsets of the avian data……………...78 4.12 Predictive success of the models based on extant bird claws……………………...... 82 4.13 Variance-covariance loadings for each variable included in the analysis…………….83 4.14 Posterior probabilities for fossil taxa generated by both predictive models…...….85-87

List of Abbreviations

μm micrometre cm centimetre ANOVA Analysis of Variance BMNHC Beijing Museum of Natural History CAGS Chinese Academy of Geological Sciences CLMB Climbing D-I Digit one D-II Digit two D-III Digit three D-IV Digit four DNHM Dalian Natural History Museum, Dalian, Liaoning, China EME Transylvanian Museum Society in Cluj FMNH The Field Museum of Natural History FRDC-GS Fossil Research and Development Center (FRDC), Third Geology and Mineral Resources Exploration Academy, Gansu Provincial Bureau of Geo-Exploration and Mineral Development in Lanzhou, China GRND Ground-dwelling HG Paleontological Center, Bohai University, Jinzhou City, China HGM Henan Geological Museum, Zhengzhou, China IS Inner Sheath arc: Angle of the arc of drawn over the ventral surface of the claw sheath from toe pad to sheath tip. IS2 Inner Sheath arc (modified): Angle of the arc drawn over the ventral surface of the claw sheath, not including the toe pad. IU Inner Ungual arc: Angle of the arc drawn over the ventral surface of the ungual bone. IQR Interquartile Range IVPP Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences (CAS) JPM Jinzhou Paleontological Museum LDA Linear Discriminant Analysis

7 ln Natural logarithm LPM Liaoning Palaeontological Museum m metre mm millimetre MANCH Manchester Museum MB Museum Berlin (see MFN) MFN Museum für Naturkunde Berlin (Berlin Natural History Museum) MPC Institute of Paleontology and Geology, Mongolian Academy of Sciences, Ulaanbaatar, Mongolia NA Not applicable NML National Museums of Liverpool NMS National Museums of Scotland OS Outer Sheath arc: Angle of the arc drawn over the dorsal surface of the claw sheath. OU Outer Ungual arc: Angle of the arc drawn over the dorsal surface of the ungual bone. PBMS Predatory Bird Monitoring Scheme, the Lancaster Centre for Ecology & Hydrology (CEH) PERMANOVA Permutational multivariate analysis of variance PKUP Geological Museum of Peking University PRCH Perching PRED Predatory STM Shandong Tianyu Museum of Natural History UMNH Natural History Museum of Utah YFGP Yizhou Fossil & Geology Park ZPAL Institute of Paleobiology, Warsaw

Abstract Claws are involved in numerous vital functions including locomotion and prey capture, and as a result, animals are expected to evolve claw morphologies that enable efficient performance of these biological roles whilst minimising stress/strain during functional performance. This is supported by past studies finding the geometry of pedal claws to correlate with mode of life for extant birds, squamates, and mammals, and this relationship has frequently been cited to infer lifestyles of Mesozoic theropods including Archaeopteryx and Microraptor. However, claw sheaths are comprised of soft tissues that rarely fossilise and are prone to breakage and deformation, and so many fossil claws are comprised solely of the internal ungual bone, and even fossilised claw sheaths from exceptionally preserved specimens tend to be broken and/or deformed. Past inferences based on intact extant claws are thus compromised by errors related to reconstruction and/or comparison of nonhomologous structures (claw sheaths and ungual bones). As the ungual phalanx within the claw is more commonly preserved intact in the fossil record, the geometry of this bone may provide a more useful metric for palaeontological analysis. In this study, ungual bones of the third pedal digit belonging to 95 species of extant birds representing 24 orders and 5 species of extant squamates have been radiographed using the portable dental imaging device the Nomad Pro Radiography Unit. Curvatures of the ventral and dorsal surface of this bone were measured by approximating the surface to a circular arc and taking degree of the arc using a custom-made software DinoLino.exe. A predictive model created using linear discriminant analysis (LDA) found a significant relationship between these measures on the ungual bone and four modes of life; ground- dwelling, perching, predatory, and climbing; with total weighted accuracy equal to 0.79 when tested on extant birds. Ranges and medians of the measured claw angles were relatively low for ground-dwelling taxa, relatively high for climbing taxa, and relatively intermediate for perching and predatory taxa; these results suggest ungual bones follow similar trends as keratinous sheaths with regard to ecomorphology. However, there is much overlap between ranges found for claw angle and also between behavioural categories. The predictive model predicts perching ecologies for Archaeopteryx, Balaur, Xiaotingia, Zhenyuanlong, and Microraptor, a predatory ecology for Confuciusornis, Anchiornis, Changyuraptor, and Talos, a ground-dwelling ecology for Borogovia, Eosinopteryx, and Halszkraptor, and a climbing ecology for Sapeornis. Some predictions were unexpected, but predictions were generally consistent with fossil evidence. Unfortunately, many fossil claws measured here possess morphologies that are intermediate relative to those of extant taxa and as a result many predictions by the model are based on low posterior probabilities. The results suggest ungual bone geometry is a useful metric that could shed light on debates in palaeontology including the theropod-bird transition, and the evolution of avian flight. However, further work could be useful in mitigating the confounding influences of scaling and phylogeny, investigating whether overlapping morphospaces may be distinguished, and improving strength of classifications.

Declaration

No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or institute of learning.

Copyright statement i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and she has given the University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=24420), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy on Presentation of Theses

Acknowledgements

I’d like to thank Dr. Bill Sellers and Dr. Andrew Chamberlain for their supervision. Special thanks to Callum McLean for manuscript revisions and support.

Thank you to my family, particularly my parents, who have always been supportive, my older brothers, and my grandparents, who made this possible with their financial support.

I’d also like to thank Tony Parker from the World Museum of Liverpool, Rachel Petts from The Manchester Museum, Bob McGowan from The National Museum of Scotland at Edinburgh, and Elaine Potter of the Predatory Bird Monitoring Scheme in Lancaster CEH for their contribution of extant specimens for this study. Thank you to all institutions from which I sourced fossil specimens to analyse in this study.

1. Literature review

1.1 Avians, avialans, and theropod dinosaurs

Modern birds are the direct descendants of non-avian theropod dinosaurs, specifically maniraptoran coelurosaurs. This is supported by an overwhelming amount of evidence including similar reproductive traits (Zelenitsky, 2006), growth patterns and metabolic rates

(Bakker, 1972, 1986; Erickson et al., 2009), integumentary features (Norell and Xu, 2005;

Turner, Makovicky and Norell, 2007), genomic size (Organ et al., 2007), osteological similarities (Ostrom, 1976), taxa such as Archaeopteryx (Ostrom, 1976, 1985) and Jeholornis

(Zhou and Zhang, 2003) that possess dinosaurian and avian characters (e.g. long bony tails and primitive, clawed wings), and several small, feathered coelurosaurs discovered in China

(Xu, Zhou and Wang, 2000; Czerkas and Yuan, 2002; Hu et al., 2009; Xu et al., 2009, 2011,

2014; Agnolín and Novas, 2013; Han et al., 2014; Sullivan et al., 2014). Analyses of extant birds thus have potential to provide insights regarding non-avian theropod dinosaurs, particularly maniraptorans, and other extinct animals in the avian evolutionary lineage.

Though the ancestral relationship between birds and dinosaurs is widely accepted, the manner in which ground-dwelling theropods developed the capability for powered flight, and the evolutionary drivers behind it, are still widely debated. Initially, a dichotomy existed between the ‘ground-up’ hypothesis of flight evolution, which suggests flight evolved when some cursorial, non-avian theropods evolved flight capabilities as an extension of leaping from a running start (Ostrom, 1974, 1976, 1985, 1986; Padian, 1982; Padian and Chiappe,

1998; Burgers and Chiappe, 1999), and the ‘trees-down’ hypothesis of flight evolution, which suggests that the ancestors of birds transitioned through an arboreal gliding stage prior to evolving powered flight (Marsh, 1880; Bock, 1965, 1983; Parkes, 1966; Norberg, 1985).

Fossil evidence supports the ‘tree-down’ hypothesis, with numerous small coelurosaurs and primitive birds that possess putative arboreal adaptations having been discovered in Jehol, China (Xu, Zhou and Wang, 2000; Czerkas and Yuan, 2002; Chatterjee and Templin, 2004; Hu et al., 2009; Wang, O’Connor and Zhou, 2014; Hu, O’Connor and

Zhou, 2015). These taxa appear to represent an intermediate arboreal stage between cursorial theropods and Aves. Biomechanical studies also suggest that developing flight through arboreal gliding is more likely, as gliding from a perch requires far less power than taking off from the ground (Norberg, 1985; Chatterjee and Templin, 2004). An arboreal gliding stage also provides an explanation for gradual stepwise evolution of the wing and flight apparatuses seen in the fossil record, as gliding only requires some of the adaptations that are a prerequisite for powered flight (Marsh, 1880; Bock, 1965, 1983).

Biomechanical models have supported gliding ability for Microraptor, which has extensive forelimb and hindlimb feathering creating what has been called a tetrapterygid, or

“four-winged” gliding style (Chatterjee and Templin, 2007; Alexander et al., 2010). The bizarre scansoriopterygids have also been interpreted as being gliding-capable based on their extremely long forelimbs as well as membranous tissues interpreted as a patagium, or membranous wing, in Yi and Ambopteryx fossils (Czerkas and Yuan, 2002; Zhang et al., 2008;

Wang et al., 2019). The scansoriopterygid foot is interpreted as being adapted for arboreality based on phalangeal proportions (Czerkas and Yuan, 2002), and Microraptor has frequently been interpreted as being scansorial (i.e. capable of arboreal locomotion) or an arboreal specialist based on certain features (e.g. elongate forelimbs, recurved claws) (Xu, Zhou and

Wang, 2000; Burnham et al., 2011). The apparent arboreal and gliding adaptations of these and other Mesozoic theropods (microraptorines, Archaeopteryx) support a phugoid gliding ecology in which derived maniraptorans and avialans glided between branches (Chatterjee and

Templin, 2007). 14

Figure 1.1: Maniraptoran taxa with apparent gliding adaptations. (A) A reconstruction of

Microraptor (Li et al., 2012) (B) A diagram of Yi qi with membranous wings extended (Xu et al., 2015)

Maniraptoran taxa with relatively elongate forelimbs (Microraptor, Anchiornis, Yi,

Epidexipteryx, Xiaotingia), extremely curved claws on the manus and/or pes that would have enabled claw-based climbing (Microraptor, Yi, Epidexipteryx), very small sizes (Microraptor,

Anchiornis, Xiaotingia, Yi, Epidexipteryx), extensive feathering on the hindlimbs that may have hindered terrestrial locomotion (Pedopenna, Microraptor, Anchiornis, Changyuraptor)

(Xu et al., 2003; Alexander et al., 2010; Han et al., 2014), and/or purported gliding ability

(Microraptor, Yi, Epidexipteryx) have been interpreted as arboreal specialists. All maniraptorans possess features that might be adaptive for arboreality including elongated forelimbs capable of supinating in unison, a powerful shoulder girdle as indicated by the biceps tubercle and ossified sternum, laterally facing glenoids to enable dorsoventral humeral movement, inward-facing palms, elongate phalanges, and curved claws (Naish, 2000b;

Chatterjee, 2015). These characters indicate that climbing style for non-avian theropods would 15 have involved grasping the tree, digging the claws into the trunk, and pulling with the forelimbs while “walking” up the trunk with the hindlimbs in the fashion of a wolverine (Gulo gulo) (Figure 1.2) (Naish, 2000b).

Figure 1.2: Proposed mode of climbing for theropod dinosaurs. (Left) a wolverine (Gulo gulo) climbing a tree trunk (Naish, 2000b) (Right) a maniraptoran climbing a tree trunk in the same fashion. Climbing maniraptoran drawn in Inkscape and loosely based on Microraptor.

All coelurosaurs possess structures that indicate enhanced stereoscopic vision and large brain size, both of which are adaptive for foraging and locomotion in visually and spatially complex arboreal environments (Chatterjee and Templin, 2004; Chatterjee, 2015). It is possible that the large brains and enhanced vision in coelurosaurs initially evolved in response to predatory lifestyles (Stevens, 2006; Witmer and Ridgely, 2009) and that these 16 adaptations could have aided the transition to arboreal habitats by more derived taxa.

Evolutionary increasing of flexibility of wrists and hands is also observed within

Coelurosauria, and it has been suggested that this occurred is in response to increasing arboreal habits to enable grasping of branches and tree trunks (Chatterjee, 1997).

Coelurosaurs, particularly dromaeosaurs, underwent extreme miniaturisation prior to the evolution of flight. Fossil avians, avialans, and many derived maniraptoran taxa have small body size (<4 kg) that make them a candidate for arboreality, particularly dromaeosaurs (e.g.

Microraptor, Changyuraptor, Zhongjianosaurus) but also scansoriopterygids and other maniraptorans (Anchiornis, Xiaotingia) (Xu, Zhou and Wang, 2000; Carrano, 2006; Turner et al., 2007). Small size is adaptive for arboreal animals because it lessens impact upon falls from tree branches (Olshevsky, 1994; Naish, 2000a), the energy requirements during climbing are less for a small animal, and smaller animals may be supported by more and smaller branches (Naish, 2000a).

Interpretations of arboreality in non-avian theropods have been hotly contested. Some workers have suggested that certain features interpreted as arboreal (e.g. sharply recurved claws) only signify phylogenetic signal and adaptations to a predatory lifestyle (Chiappe,

1997, 2001; Padian and Chiappe, 1998). Other workers have speculated that coelurosaurs with apparent arboreal adaptations could have been facultatively arboreal but no non-avian theropod could have been an arboreal specialist because they lack certain prehensile tails and reversible ankles, which have convergently evolved in many lineages of arboreal marsupials, eutherians, and squamates (Naish, 2000a; Dececchi and Larsson, 2011). However, many extant arboreal taxa including birds, carnivorans, and squirrels also lack prehensile tails

(Cartmill, 1985), and some taxa are adept climbers (e.g. goats, hyraxes) despite possessing no obvious adaptations for climbing (Dececchi and Larsson, 2011).

Recent observations from juvenile birds, seen as a proxy for transitional avian taxa, have led workers to suggest an alternate hypotheses of the evolution of flight and flapping in theropod dinosaurs (Dial, 2003; Dial, Randall and Dial, 2006; Dial, Jackson and Segre, 2008).

One hypothesis that has gained support is Wing-Assisted-Incline-Running (WAIR), which suggests that non-avian theropods could have developed flapping motions and primitive wings to boost power for running up inclined surfaces. Work on juvenile chukar partridges has demonstrated that they boost power from flapping their undeveloped wings when running up inclined tree trunks (Dial, 2003). It has been suggested that this behaviour could have provided a pathway for the development of proto-wings and rudimentary flapping behaviours in non-avian theropods. Although this provides an explanation of how mainly cursorial non- avian theropods could have developed into birds capable of powered flight, the theory does not explain why stem birds with arboreal adaptations are seen in the fossil record. Other workers have suggested that flight may have developed by “hang-gliding” off cliffs from a running jump (Norberg, 2007); however, there are few examples of birds using cliffs specifically to take off in modern birds and this theory again fails to explain the onset of arboreal adaptations in early birds.

1.2 Comparative study and claws

Direct evidence of fossil behaviours and ecology (e.g. stomach contents) is rare. A few exceptional fossils preserve animals engaged in a behaviour, but for the majority of fossil taxa lifestyle must be inferred based on anatomical features. Though some forms are maladaptive or exapted, most organisms possess structures that are adapted for performing essential functions (Biewener, 1983). Biological form thus tends to reflect function and, subsequently, lifestyle. This concept underlies comparative anatomy, in which a relationship is measured

18 between anatomical forms and functions and/or ecologies for extant taxa and then used to infer function and/or ecology for analogous fossil taxa (Kardong, 2012).

Claws are a popular metric in comparative study for their involvement during locomotion and ecological functions (e.g. prey capture), reported comparability across taxa, and relatively simple geometry. Predatory, perching, and climbing/scansorial taxa are expected to possess relatively curved claws to improve grip (Petie and Muller, 2007; Tulli,

Abdala and Cruz, 2011), and ground-dwelling taxa are expected to possess claws that lie flat against the substrate during locomotion. This has been substantiated by numerous studies reporting claw curvature increases with degree of arboreality and predatory habits for a diverse group of amniote taxa including mammals (Van Valkenburgh, 1987; MacLeod and

Rose, 1993; Burnham et al., 2011), squamates (Zani, 2000; Birn-Jeffery et al., 2012; Crandell et al., 2014), and birds (Feduccia, 1993; Pike and Maitland, 2004; Glen and Bennett, 2007).

As this relationship appears to be universal among amniotes, degree of claw curvature has been frequently cited as evidence for ecology and lifestyle in Mesozoic theropods and other fossil taxa (Feduccia, 1993; Spielmann, Renesto and Lucas, 2006; Burnham et al., 2011).

1.2.1 Claws, locomotion, and ecology

Generally, flatter claws are observed among ground-dwelling taxa, moderately curved claws are observed among predatory taxa and perching birds, and sharply curved claws are observed among climbing/scansorial taxa (Feduccia, 1993; Zani, 2000; Pike and Maitland,

2004). The relation between claw curvature and lifestyle is thought to result from the evolutionary reduction of bending stresses during terrestrial locomotion through claw flattening (Alexander, 1968, 1989; Biewener, 1983; Biewener and Patek, 2018) and, conversely, the evolutionary increase of claw curvatures to increase grip ability for enabling

19 climbing, prey capture, and perching functions (Pike and Maitland, 2004; Petie and Muller,

2007).

Transverse forces leading to bending moments are the primary generator of stress during locomotion because long, slender structures such as bones and claws are more likely to break from bending than from axial compression (Rubin and Lanyon, 1982; Biewener, 1983;

Alexander, 2003). Minimisation of bending moments is thus a common evolutionary tactic for reducing stress loading upon structures. Pedal claws are compressed under the weight of the animal during terrestrial locomotion, which can cause bending. Assuming that claws function as simple beams, flatter claws will experience reduced bending and increased tangential stresses during terrestrial locomotion compared to curved claws (Gere and Timoshenko,

1985). As tangential stresses do not threaten structural integrity (Alexander, 2003), flat claws are adaptive for terrestrial locomotion. Claw flatness is expected to correlate with terrestriality/cursoriality, but ground-dwelling taxa may still possess curved claws. This will stem from genetic factors or from claw multifunctionality, which necessitates an evolutionary compromise (Marshall, 2006) between claw flattening to reduce stress and curving to enable certain functions.

For taxa that rely upon claws whilst climbing, it is hypothesized that curved claws enable clinging by interlocking with the substrate to create a new, nonvertical contact surface between the animal and the vertical surface (Cartmill, 1985). Numerous observational studies have found that high claw curvatures correlate with scansoriality across unrelated taxa as previously mentioned, and increased claw curvature has been reported to correlate with enhanced clinging and climbing performance in lizards (Zani, 2000; Tulli, Abdala and Cruz,

2011; Crandell et al., 2014). As a result, fossil taxa with very sharply curved claws (e.g.

Microraptor) are often interpreted as scansorial or arboreal (Xu, Zhou and Wang, 2000;

Czerkas and Yuan, 2002; Wang, O’Connor and Zhou, 2014). However, biomechanical 20 significance of claw curvature is not well understood (see Zani, 2000). Further research is necessary regarding clade-specific ecomorphologies and the functional significance of degree of claw curvature.

Physical models based on the claws of predatory insects found that curved claws are more efficient than straight claws at holding prey because they reduce forces that might drive prey away from the pivoting point, increase resulting forces that drive prey toward a pivoting point for stabilisation, and enable perfect clamping (Petie and Muller, 2007). Though it is clear that curved claws are useful for taxa with a grasping function for the manus and/or pes

(e.g. perching, predation, climbing), it is unclear whether functions correlate with distinct ranges of claw curvature. Some past studies have reported that claw curvatures overlap significantly between predatory birds and non-raptorial perching birds (Pike and Maitland,

2004; Csermely and Rossi, 2006; Birn-Jeffery et al., 2012). It is plausible that grasping branches could influence the evolution of similarly moderate claw curvatures as grasping prey, but other studies have reported distinct claw morphologies not only for non-raptorial perching birds and raptors but also for individual raptorial groups (Fowler, Freedman and

Scannella, 2009; Csermely, Rossi and Nasi, 2012; Tsang et al., 2019).

Multiple studies have reported claws of owls, accipiters, falcons, and non-raptorial perching birds group separately (Csermely and Rossi, 2006; Fowler, Freedman and Scannella,

2009; Csermely, Rossi and Nasi, 2012; Tsang et al., 2019). This is thought to relate to different prey-type specialisations and killing strategies of each group. Distinct claw morphologies appear to correlate with distinct grip abilities: in vivo studies reported that raptors produce greater grip forces and falcons produce greater bite forces, which is to be expected as falcons kill with their beaks and accipiters kill by squeezing with the claws

(Sustaita and Hertel, 2010). These studies report that non-raptorial perching birds have less curved claws than raptors, but relationships between form and function for raptorial groups 21 seem to differ based on metrics utilised. Falcons, for example, are reported to possess intermediate claw morphologies and low (dorsal) claw curvatures by some studies (Csermely and Rossi, 2006; Csermely, Rossi and Nasi, 2012) and the highest measured (ventral) claw curvatures of all predatory birds by another (Fowler, Freedman and Scannella, 2009).

1.2.2 Quantifying claw form

Methods for quantifying claw curvature vary widely due to conflicting assumptions regarding claw shape, relevance of various features, locations of landmarks, and appropriate morphometric techniques. Claw form has been quantified using outlines (MacLeod and Rose,

1993; Lautenschlager, 2014), landmarks (Tinius and Russell, 2017; Tsang et al., 2019), and traditional morphometric approaches (e.g. linear measurements) (see Tinius and Russell,

2017). It has been suggested that claws follow the shape of logarithmic spirals (Thompson,

1942; Mattheck and Reuss, 1991), parabolic curves (Peters and Görgner, 1992), or circles

(Feduccia, 1993; Pike and Maitland, 2004; Petie and Muller, 2007) (Figure 1.3). Others have quantified claw form using geometric abstractions that make no assumption of shape (Zani,

2000) (Figure 1.3). Though these approximations do not entirely capture claw form and its complexity, a recent meta-analysis reported that all geometric methods except logarithmic spirals capture the bulk of variation in avian claws and correlate significantly with mode of life (Tinius and Russell, 2017).

Figure 1.3: Methods of quantifying claw curvature. (Tinius and Russell, 2017)

A. A Picoides villosus claw measured using a parabolic arc drawn from claw tip to middle

of the claw base after Peters and Görgner (1992)

B. A Picoides villosus claw measured using pseudolandmarks (Tinius and Russell, 2017)

C. A Picoides villosus claw measured using triangles drawn from landmarks on the claw

representing the claw’s base, tip, and inflection point if present or midpoint of AB after

Zani (2000)

D. A Picoides villosus claw measured with a circle drawn from three landmarks on base

of the toe pad, tip of the claw, and midpoint of AB after Feduccia (1993)

E. A Picus canus claw measured using a logarithmic spiral as proposed by Thompson

(1942)

F. A Picoides villosus claw measured using two best-fit circles as employed by Petie and

Muller (2007) and others

Approximating the claw to the arc of a circle is the most popular method of quantifying curvature. Claws have been compared to circular arcs in two primary ways: by matching best- fit circles to the ventral or dorsal arc of the claw, and by drawing a circle from three landmarks taken on the claw. Curvature has been measured as the inverse of the radius (Petie and Muller,

2007) and/or as an arc of the circle (Feduccia, 1993; Pike and Maitland, 2004; Glen and

Bennett, 2007).

Feduccia (1993) established a popular method for measuring claw curvature by placing three landmarks A at the tip of the claw, B at the base of the toe pad, and X at the intersection of line AB̅̅̅̅ with the underside of the keratinous sheath, using these points to draw a circle for which the ventral claw surface is an arc (Fenn, 2001), and taking claw angle for the arc

(Figure 1.4). Pike and Maitland (2004) utilised a similar method to measure outer curvature of the claw by shifting landmark A to the base of the dorsal surface of the claw sheath (Figure

1.4 B). The authors suggest that this metric is more reliable than ventral curvatures measured after Feduccia (1993) because the dorsal arc of the claw tends to more closely approximate a circular arc.

Figure 1.4: Methods of calculating claw arc.

A. The “Feduccia” method of measuring ventral claw arc (Feduccia, 1993). A line is

drawn between the claw tip (A) and the base of the toe pad (B), a chord bisector DC̅̅̅̅ is

drawn perpendicularly from A̅̅̅B̅, a point (X) is drawn at the intersection of DC̅̅̅̅ and the

ventral surface of the keratinous sheath, at the intersection of perpendicular bisectors

to chords AX̅̅̅̅ and BX̅̅̅̅ the centre E′ of a hypothetical circle is drawn, and claw angle is

measured between radii E′A and E′B.

B. The “Pike and Maitland” method of measuring dorsal claw arc. A line is drawn

between the claw tip (A) and the junction of dorsal sheath and skin (A′), a chord

bisector BB′̅̅̅̅̅ is drawn perpendicularly from AA′̅̅̅̅̅, a point (C) is drawn at the intersection

of BB′̅̅̅̅̅ and the dorsal surface of the keratinous sheath, at the intersection of

perpendicular bisectors to chords AC̅̅̅̅ and ̅A̅̅′̅C̅ the centre O of the hypothetical circle is

drawn, and claw angle is measured between radii OA′ and OA.

Some workers favour inner curvature of the claw because this surface directly interfaces with the substrate during functional performance and as such it is more biomechanically relevant and likely to greater reflect adaptations (Feduccia, 1999; Burnham et al., 2011; Tinius and Russell, 2017). Other workers favour outer curvature of the claw because it more closely approximates the arc of a circle and is less subject to variation resulting from size and position of flexor tubercles and toe pads (Pike and Maitland, 2004; Csermely and

Rossi, 2006; Tinius and Russell, 2017). Recent workers have utilised both inner and outer curvatures of the claw and reported both provide useful and subtly different information

(Fowler, Freedman and Scannella, 2009; Birn-Jeffery et al., 2012; Patiño and Fariña, 2017). A pseudolandmarks analysis on avian pedal claws indicated that data for dorsal and ventral arcs follow a similar distribution, but dorsal claw arcs tend to more closely approximate a circular

25 arc whereas ventral surfaces are more complex (Tinius and Russell, 2017). The ventral surface of the claw is probably specialised for interacting with substrates whereas the dorsal arc is a generalised curve that provides structural support.

Different features of the claw (e.g. toe pads, claw tips) have been included in past measurements based on their inferred functional relevance, variability, and preservation potential. Measures including the toe pad are more influenced by soft tissue variability and less applicable to fossil claws, which rarely preserve soft tissues. However, enlarged toe pads correlate with raptorial lifestyles in birds (Tsang, Wilson and McDonald, 2019) and are known to provide enhanced grip ability for arboreal locomotion in lizards (Crandell et al., 2014) and mammals (Cartmill, 1974). Fowler and colleagues (2009) omitted toe pads in measures of inner claw curvature to improve comparability with fossil taxa; a meta-analysis later reported that this measure displays a significant but weaker correlation with avian lifestyle than the equivalent measure including toe pads (Tinius and Russell, 2017). Glen and Bennett (2007) excluded claw tips in their measures of curvature on the basis that claw tips are highly variable, but this is likely to erase ecological signal as claw tips contact the substrate during locomotion and are functionally relevant.

Unfortunately, the tip of the keratinous sheath is greatly influenced by physical wear that can change phenotype (Davies, 1987; Juanes and Hartwick, 1990). Amniote claws are comprised of inner bony core, which is called the ungual phalanx or ungual bone, and an external keratinous sheath (Hamrick, 2001). Measures taken on the ungual bone within the claw (Figure 1.5) do not contact the substrate and thus phenotype cannot be distorted by physical wear. In addition, size, position, and shape of the flexor tubercle at the base of the ungual bone may provide a proxy for grip and ecology because it is the attachment site for flexor tendons (MacLeod and Rose, 1993; Fowler, Freedman and Scannella, 2009; Tsang et al., 2019). Therefore, measures including the flexor tubercle of the ungual bone may provide 26 similar information to measures taken on toe pads without being negatively impacted by soft tissue variability or low preservation potential.

Figure 1.5: Internal structure of the claw. Radiograph of a claw from Crinifer zonurus, specimen 6265 from the Liverpool World Museum. Radiograph taken at .20 exposure with the

Nomad Pro Radiography Unit and edited in the SIDEXIS software. The darker grey, outer structure is the keratinous sheath, the off-white, inner structure is the ungual bone, the node at the proximal end of the ungual phalanx is the flexor tubercle to which tendons attach, and the toe pad is the fleshy base encasing the flexor tubercle.

1.2.3 Quantifying behaviour

Quantifying behaviour presents some challenges. In his seminal study on claw ecomorphology, Feduccia (1993) correlated claw angles with behavioural categories ‘ground- dwelling’, ‘perching’, and ‘climbing’ for extant birds. This is a gross oversimplification of behavioural complexity, but the issue is difficult to resolve as modern birds are extremely

27 diverse and categorisation is necessary for most analyses. Some workers have attempted to account for the often intermediate nature of behaviour by classifying taxa along a spectrum of arboreality (Spielmann, Renesto and Lucas, 2006; Glen and Bennett, 2007). Other workers have included a predatory group in additional to the traditional locomotor groups (Pike and

Maitland, 2004; Birn-Jeffery et al., 2012). Inclusion of a predatory group better captures behavioural diversity of modern birds and, presumably, Mesozoic theropods.

Inclusion of predatory talons can give analyses potential to resolve whether curved claws of some non-avian theropods had a predatory or an arboreal function (Chiappe, 1997,

2001; Padian and Chiappe, 1998). Hypotheses of arboreality for Mesozoic non-avian theropods and Archaeopteryx have been disputed on the basis that all predatory, non-avian theropods possess phylogenetically curved claws (Chiappe, 1997, 2001; Padian and Chiappe,

1998), and the predatory function of the dromaeosaurid “killing claw” has recently been disputed in favour of a clinging/climbing interpretation (Manning et al., 2005, 2009; Fowler et al., 2011). In light of these ongoing debates, including a “predatory” group should be considered crucial for analyses seeking to infer lifestyles for Mesozoic theropods based on their claws. However, claw curvatures of predatory birds were reported to significantly overlap with those of perching birds by multiple studies (Pike and Maitland, 2004; Csermely and Rossi, 2006; Birn-Jeffery et al., 2012), and so it is unclear whether a predatory function may be inferred by claw curvature(s) alone.

1.2.4 Comparability across taxa and structures

Comparative study relies on structures being analogous through functional analogy, biological homology, or, ideally both. However, many studies compare distantly related taxa

(e.g. mammals and theropod dinosaurs), taxa and structures that are biomechanically dissimilar (e.g. quadrupeds and bipeds), and/or nonhomologous structures (e.g. claw sheaths 28 and ungual bones). Metric of comparison must be constrained to functional analogues and/or biological homologues in order to yield meaningful results (Kardong, 2012). However, comparisons have been made between nonhomologous structures including manual claws and pedal claws (Feduccia, 1993; Spielmann, Heckert and Lucas, 2005; Spielmann, Renesto and

Lucas, 2006; Burnham et al., 2011), claws belonging to different digits (Van Valkenburgh,

1987; Lautenschlager, 2014), and claw sheaths and ungual bones (Birn-Jeffery et al., 2012).

Compositional and development differences between claws of different amniote groups (Hamrick, 2003; Alibardi, 2009; Greenwold et al., 2014) may influence the evolution of different functional morphologies in response to similar selective pressures. Claws of modern avians and reptiles are comprised of alpha keratins and relatively tough beta keratins, whereas mammalian claws are primarily comprised of alpha keratins (Greenwold et al., 2014;

Liu, Zhang and Ritchie, 2018). Mammalian claws thus provide a poor comparative basis for dinosaurian claws despite their use as an analogue in past studies (Burnham et al., 2011;

Lautenschlager, 2014). Though squamate claws have similar composition to avian claws, these too may be incomparable to avians and, presumably, Mesozoic theropods based on developmental differences (Alibardi, 2008, 2009).

It has been inferred based on extant phylogenetic bracketing (EPB) (Witmer and

Thomason, 1995) that theropod dinosaurs possessed claws with similar compositions and consequently similar biomechanical properties to extant birds (Manning et al., 2005, 2009).

By the same logic, the claws of Mesozoic theropods are likely to have been developmentally similar to those of extant birds. Though few extant animals are biomechanically similar to theropod dinosaurs, as directly ancestral bipedal animals sharing many osteological features, birds may be considered more similar than squamates or mammals despite lacking a long bony tail or grasping hands. If these assumptions of comparability between birds and theropod dinosaurs are true, then it seems likely that theropod dinosaurs would have evolved similar 29 claw forms to extant birds in response to similar selective pressures. The claws of extant birds are thus more likely to provide a close analogue for theropod dinosaur claws compared to the claws of more distantly related taxa.

Some workers have compared manual claws of Mesozoic theropods (Feduccia, 1993;

Burnham et al., 2011) and other fossil reptiles (Spielmann, Heckert and Lucas, 2005;

Spielmann, Renesto and Lucas, 2006) to pedal claws of birds. It could be argued that pedal and manual claws are functionally analogous for the quadrupedal archosauromorph and drepanosaurids analysed by Spielmann and colleagues (2005, 2006). However, pedal claws and manual claws are neither biologically homologous nor functionally analogous for bipedal taxa. Regardless of whether both structures were utilised in climbing and/or predatory functions, the pedal claws of modern birds will be adapted for locomotion and the manual claws of Mesozoic theropods will not, and so inferences based on their comparison are flawed.

Study must be constrained to one or more homologous digits to ensure comparability.

Empirical evidence suggests significant interdigital variation between claw and toe morphologies in extant birds (Fowler, Freedman and Scannella, 2009). Past authors often constrain digit of study to the central, longest digit, which is digit III (D-III) for many birds, squamates, and Mesozoic theropods. This digit is functionally significant as the primary weight-bearing digit and the first and last point of contact with the substrate during terrestrial locomotion (Peters and Görgner, 1992; MacLeod and Rose, 1993; Tulli et al., 2009). D-III claw morphology is thus expected to be well-adapted for locomotion and other functions.

Studies reporting that D-III claw curvatures correlate with avian and squamate lifestyles support this (Feduccia, 1993; Pike and Maitland, 2004; Birn-Jeffery et al., 2012). However,

D-III is not the longest digit for all taxa; in many lizards, D-IV is the longest (Crandell et al.,

2014), and so D-III may not represent a functional analogue across vertebrate taxa.

1.2.5 Internal structure of the claw

Keratins are softer than cortical bone and are thus more prone to breakage, deformation, and decay and less prone to fossilisation (Rothschild et al., 2013). Well- preserved ungual bones are thus more common than keratinous sheaths in the fossil record, and so many inferences for fossil claw functions are based on fossilised ungual bones. This is problematic if these inferences are based on a relationship found for extant claw sheath and/or toe morphology. Ungual bones are neither functionally nor biologically comparable to claw sheaths, and so studies that compare fossil ungual bones to extant claw sheaths are flawed

(Birn-Jeffery et al., 2012). It is important to consider that an ungual bone must be less curved than the claw sheath to fit inside it as has been observed for some modern birds (Glen and

Bennett, 2007; Fowler, Freedman and Scannella, 2009). Naturally low curvatures of ungual bones may have influenced false predictions of ground-dwelling lifestyle in past studies that compare fossil ungual bones to extant claw sheaths (Spielmann, Renesto and Lucas, 2006;

Glen and Bennett, 2007; Birn-Jeffery et al., 2012).

Internal structure of the claw has not been sufficiently studied. Bending moments are expected to similarly affect both structures, and the bony core cannot drastically differ in shape if it is to fit inside the sheath. It thus seems likely that phenotypes for ungual bones follow similar trends compared to phenotypes of claw sheaths. This is supported by studies on mammalian ungual bones reporting that ungual bones roughly mirror shape of the external claw (Van Valkenburgh, 1987; MacLeod and Rose, 1993; Patiño and Fariña, 2017). However, no comprehensive analysis has measured ungual bone ecomorphology for avians or squamates.

Past studies indicate that ungual bone morphology possesses some significant correlation with mode of life, but details of this correlation are unclear. Van Valkenburgh

(1987) reported that arboreal and predatory carnivorans possess more curved ungual bones relative to ground-dwelling carnivorans based on averaged linear measurements between claws of different digits for 61 species of extant mammals. MacLeod and Rose (1993) reported similar results for a phylogenetically broader sample of extant mammals based on outline analysis and similar linear measures on D-III. More recent findings on ungual bones are unreliable; one study combined measures on sheaths and ungual bones, which are nonhomologous structures, into a single dataset for analysis (Patiño and Fariña, 2017), and another was unable to constrain digit of measure and compared distantly related taxa

(Lautenschlager, 2014).

The function of the ungual bone may be subtly different from the function of the keratinous sheath. Though ungual bones are involved in the same behaviours, they are not directly in contact with the substrate. The ungual bone is subjected to different loading patterns and magnitudes related to discrepancies in size, position, and composition. Given the known significance of stress and strain in the evolution of functional morphology, it is possible that such discrepancies could result in a significantly different relationship between form and mode of life for the ungual bone compared to the keratinous sheath.

Measurements of fossil claw angle are often taken on reconstructed claw sheaths (Glen and Bennett, 2007) or taken directly on ungual bones (Spielmann, Renesto and Lucas, 2006;

Birn-Jeffery et al., 2012). Past attempts to find a relationship between curvature of the keratinous sheath and the ungual bone have yielded variable results (Yalden, 1985; Glen and

Bennett, 2007; Patiño and Fariña, 2017). Patiño and Fariña (2017) measured ratios of 0.45,

0.75, and 0.85 between claw sheath and ungual bone curvatures for three Pleistocene ground sloths, which suggests any relationship determined is likely to have high variance. Another study reported a linear relationship between curvature of the bone and of the sheath, but the authors provided little information regarding the relationship or the sample for which it was 32 measured and utilised an unusual measurement that excludes the tip of the claw (Glen and

Bennett, 2007). Thus, at present reconstructed claw angles are based either on supposition or on a relationship for which no consensus has been reached. Values of curvature obtained for fossil claws are thus unlikely to provide reliable information when analysed using current methodologies.

1.2.6 Techniques for quantitative analysis

Organisms may evolve different forms in response to similar selective pressures to achieve similar functions (Alfaro, Bolnick and Wainwright, 2005; Wainwright et al., 2005).

One should analyse multiple measures of biological form because divergent forms preclude a

“single, straightforward relationship between form and function”(Tsang et al., 2019).

Multivariate analysis, or a “many-to-one” model, is recommended over analyses of single phenotypic features even for a structure like the claw that appears relatively straightforward in its morphological variation. Utilising multiple variables (e.g. claw measurements) also makes possible a wide range of statistical techniques that may be better suited to addressing specific research questions (Fukunaga, 2013). However many past studies utilise a single measure of claw curvature, and findings are presented as simple box plots (Feduccia, 1993; Pike and

Maitland, 2004; Glen and Bennett, 2007).

For studies attempting to class taxa using multiple measurements, using dimensionality reduction techniques such as linear discriminant analysis (LDA) or principal components analysis (PCA) to create a predictive model can be very useful (Prasad and Bruce, 2008;

Fukunaga, 2013). Though many studies have analysed multiple measures of the claw in relation to lifestyle (Zani, 2000; Tulli et al., 2009; Birn-Jeffery et al., 2012), only a fraction have utilised dimensionality reduction tools to characterize separation between functional and/or behavioural groups (Zani, 2000; Tulli et al., 2009). Of these, many rely on PCA (Zani, 33

2000) or techniques derived from PCA (Tulli et al., 2009) despite this method being reported to have less discriminatory power than LDA.

2. Introduction

One way to overcome limitations of past studies is to study the ungual bone so that fossil claws not preserving soft tissues may be directly compared to extant claws. If form of the ungual bone proves to be a useful metric, then direct comparative analysis will be made possible for a greatly expanded dataset of fossil material. Additionally, soft tissue reconstruction, which introduces another source of error into analysis, would be made unnecessary in such study if measures taken solely on the ungual bone provide a relatively accurate, precise proxy for lifestyle. By analysing the correlation between ungual bone curvatures and mode of life, errors related to reconstruction and/or comparison of nonhomologous structures should be minimised.

This study investigates the relationship between dorsal and ventral curvatures of ungual bones and behavioural categories of ground-dwelling, perching, predatory, and climbing for a diverse group of extant avians and squamates. The results are then utilised to infer lifestyle for a sample of fossil paravians and avians. Predictive models will be created using discriminant analysis because this method provides enhanced discriminatory power compared to PCA or non-parametric techniques and thus provides a more powerful classification tool (Prasad and Bruce, 2008; Fukunaga, 2013). It is hoped that the enhanced discriminatory power of LDA will resolve issues of overlap between morphospaces reported by past studies (Pike and Maitland, 2004; Birn-Jeffery et al., 2012).

Conducting a comprehensive study on avian ungual bones is useful for multiple reasons. Claw curvature is frequently cited when inferring fossil lifestyle, but claw sheaths are 34 often broken or absent in the fossil record. The fleshy toe pad, which is often used in calculating inner curvature of the claw after Feduccia (1993), is even rarer in the fossil record.

As many fossils only preserve the inner bony core of the claw, studying ungual bone morphology will enable the direct comparison of extant and fossil material for a large sample of fossil specimens that cannot be reliably analysed using current techniques.

Though it may be less biomechanically significant, ungual bone curvature could provide a preferable metric to sheath curvatures in some ways as bone is less prone to distortion than keratin, and phenotype of this bone is unaffected by physical wear. Amniote claw sheaths are influenced by physical wear and so measurements of claw angle may not always reflect the true phenotype of the claw (Davies, 1987; Juanes and Hartwick, 1990). This presents a potential problem in particular for studies that take claw angle utilising a landmark located on the tip of the claw (Figures 1.3, 1.4). It has been suggested that keratin is constantly redeposited such that claws conform to the active phenotype and so wear at the tip has no bearing on measurements of claw angle or inferences of fossil lifestyle (Feduccia, 1999), but this claim has yet to be proven empirically.

More broadly, quantifying ungual bone geometry in this fashion expands our understanding of avian and, presumably, theropod anatomical structures and ecomorphogy.

This information could be relevant to numerous scientific enquiries including biomechanics of locomotion, perching, and predation, the evolution of ecological diversity in birds, the evolution of powered flight in birds, and more.

The inclusion of a predatory group among locomotor groupings may seem unusual, but it is not unique to this study. These categories follow those used by Pike and Maitland (2004) and Birn-Jeffery and colleagues (2012). Claw angles of raptorial birds have been found to significantly overlap with those of other groups, particularly perching birds (Pike and

Maitland, 2004; Birn-Jeffery et al., 2012), and it has been suggested that arboreal hypotheses 35 for extinct theropods are based on an erroneous interpretation of a climbing function for curved claws that in fact were curved to assist in prey capture/dispatching (Chiappe, 1997).

Distinguishing between claws with a predatory function and claws with a climbing/perching function would thus provide greater clarity to arboreal hypotheses and resolve a confounding factor. Furthermore this study seeks to determine ecomorphological categories using quantitative rather than qualitative observations.

As traditional descriptions of claw ecomorphology are not well-supported by empirical evidence (Tinius and Russell, 2017), and even less has been demonstrated regarding ecomorphology of ungual bones, qualitative features of the claw such as tapering and mediolateral constriction will not be given much weight when inferring fossil lifestyles. It may be useful for future studies to conduct an in-depth, qualitative analysis of claw ecomorphology, but this is beyond the scope of this study.

This study will focus on claw curvature as it has been reported to have a similar relationship with certain lifestyles for a phylogenetically diverse group, and it seems likely that variations in claw curvature occur due to universal biomechanical principles. Finding an ecomorphological relationship that is applicable for broad taxonomic groups would be very useful. Claw curvature will be focused on for its inferred comparability across taxa, known correlation with lifestyle, and large contribution to morphological variation in avian claws

(Tinius and Russell, 2017)

Claw curvature will be quantified as the arc of a circle drawn using three points on the claw after Feduccia (1993), Pike and Maitland (2004), and Fowler and colleagues (2009) for five reasons. First, utilising this frequently cited metric will increase comparability between these results and results of past studies (Spielmann, Heckert and Lucas, 2005; Csermely and

Rossi, 2006; Birn-Jeffery et al., 2012). Second, a recent meta-analysis reported avian claw curvatures measured using these methods correlate significantly with the behavioural groups 36 studied here (Tinius and Russell, 2017). Third, the delineation of landmarks is expected to improve precision compared to best-fit circles (Glen and Bennett, 2007; Petie and Muller,

2007) and geometric abstractions after Zani (2000), which rely on somewhat arbitrary landmarks (Zani, 2000; Ribas et al., 2004). Fourth, geometric morphometric methods are prone to overfitting the data (Hawkins, 2004), and so using linear measurements may increase comparability across taxa. Fifth, claw arc appears more pertinent to this enquiry compared to claw radius as demonstrated by Figure 2.1.

Figure 2.1: Claw radius versus claw arc. Figure drawn in Inkscape. Claws A and B have different radii but the same arc angle. Claws A, C, and D have the same radii but different arc angles. If curvature is quantified as the inverse of the radius, then claws A and B have 37 different curvatures but claws A, C, and D have the same curvature. If curvature is quantified by degree of the arc, then claws A, C, and D have different curvatures while claws A and B have the same curvature.

Figure 2.1 demonstrates that claw radius is a function of size as has been reported by some past studies (Pike and Maitland, 2004; Tinius and Russell, 2017), and so the inverse of the radius has not been utilised as a measurement of claw curvature despite this being a standard method of measuring curvature (Fenn, 2001). Variation in claw arc appears to measure changes in shape and length rather than size (Figure 2.1) and may be more ecologically relevant based on correlations reported by past studies (Feduccia, 1993; Pike and

Maitland, 2004; Birn-Jeffery et al., 2012; Tinius and Russell, 2017). Degree of claw arc is thus utilised as the measure of curvature in this study. It should be noted that this metric is related to length; elongation along the hypothetical circle increases claw arc and, subsequently, curvature as measured by this study (Figure 2.1).

Dorsal and ventral arcs will both be used for three reasons. First, it has been debated which metric is more useful, and as both sides present valid justifications, it seems reasonable to simply include both metrics in case one yields more useful information. Second, a recent meta-analysis comparing methods of analysing the claw reported dorsal and ventral arcs both had a strong correlation with the ecological group but had slightly different relationships with the ecological group and so utilising both may provide information that a single metric could not (Tinius and Russell, 2017). Third, utilising multiple quantitative variables enables the usage of more advanced statistical techniques such as discriminant analysis and improves robustness against divergent functional morphologies

Claw sheath curvatures have been included for two purposes. The primary aim is to compare values determined for ungual bones with those found for claw sheaths. It is also to test utility of a model based on both bone and sheath curvatures.

3. Materials and Methods

3.1 Development of the model

Curvatures were examined for ungual bones and sheaths belonging to 101 adult specimens of 95 species of bird representing 24 orders and 40 families, and 5 species of squamate representing 5 families. Multiple specimens were measured for some taxa. As this study seeks to infer modes of life for fossil taxa on the avian lineage, the final predictive model is based on bird claws. A small group of behaviourally diverse squamates have also been tested to investigate if determined trends are universal or constrained by clade.

Crocodilians have been excluded because modern taxa lack the behavioural diversity that this study seeks to analyse. Most sampled claws are museum specimens that were imaged on-site.

The dataset was also supplemented with specimens acquired from independent sources.

All extant specimens measured are adults to constrain potential influences of ontogenetic and/or behavioural changes during life known for some taxa (Fain and Houde,

2004). Claw curvature among digits can vary greatly (Fowler, Freedman and Scannella, 2009) and so digit of study is constrained to pedal digit III after the fashion of past studies.

Claw length was limited to a maximum of 44 mm to fit on the active sensor area and a minimum of 7 mm because fine details could not be resolved for very small claws. As body mass and claw radius correlate (Pike and Maitland, 2004), body masses for the sample taxa are limited from 36g to 1930g to constrain claw size. Body masses were determined from the

CRC Handbook of Avian Body Masses (Dunning Jr, 2009) because specimen body masses were not available. It was often not possible to sex the specimens, and in these instances body mass was calculated as the average of male and female body mass.

Each specimen was placed into one of four behavioural categories: ground-dwelling, perching predatory, or climbing. These classifications were based on the literature (Hoyo,

Elliott and Sargatal, 2001, 2002, 2004, 1992, 1996, 1999; Hoyo et al., 1994, 1997; Hoyo,

Elliott and Christie, 2003, 2005, 2006, 2008; Hoyo, Christie and Birdlife, 2007) and follow those used in past studies (Pike and Maitland, 2004; Birn-Jeffery et al., 2012). Behavioural complexity presents an issue because few animals exhibit behaviours of just one defined group. For example, galliforms, though classified as ground-dwellers, often roost in trees, and many perching birds (e.g. corvids) are comfortable locomoting on the ground. Species have been selected to best represent each group as determined by percent life spent engaging in the assigned behaviour and possession of apparent adaptations for certain lifestyles (e.g. corvids tend to walk or hop awkwardly on the ground whereas galliforms easily walk, run, and forage on the ground with regularity).

To limit confounding effects of phylogeny, extant specimens have been selected such that each behavioural category includes three or more orders and some orders are represented within multiple behavioural categories. Of the 26 climbing birds measured, over 50% are piciform birds (n=14). Climbing birds from other clades Passeriformes (n=6), Coliiformes

(n=1), and Psittaciformes (n=5) were included when possible, but most of these birds possess miniscule claws that cannot be imaged using the Nomad Pro Radiography Unit. Predatory birds within the dataset are represented by raptorial groups Falconiformes (n=4), Strigiformes

(n=9), and Accipitriformes (n=9) for their habit of seizing prey with the talons. Perching birds have a phylogenetically diverse spread with 11 orders represented, and Columbiformes (n=6)

40 represent 21.4% of the group. Ground-dwelling birds have a phylogenetically diverse spread with 13 orders represented, and galliform birds (n=8) represent 25% of the group.

Claw angles from predatory taxa overlap significantly with other groups (Pike and

Maitland, 2004; Csermely and Rossi, 2006; Birn-Jeffery et al., 2012), and so birds that utilise talons to capture and/or dispatch prey have been included to minimise likelihood of Type II error in which claws utilised for predation may be misclassified as having an arboreal function or vice versa. The predatory sample group is comprised of raptors that seize prey with the talons, but it should be noted that method of prey dispatching and prey-type specialisation vary widely within this group (Csermely and Rossi, 2006; Fowler, Freedman and Scannella,

2009; Csermely, Rossi and Nasi, 2012; Tsang et al., 2019). Finding a distinctive morphospace based on predatory claw curvatures is of interest for various enquiries and will expand the predictive capacity of the model to include ecological as well as locomotor groupings.

3.1.1 X-ray techniques

Claws were radiographed in mediolateral view using the Nomad Pro Radiography Unit

(Figure 3.1) and processed in the SIDEXIS software (https://www.dentsplysirona.com/en).

The Nomad Pro Radiography Unit is a handheld, portable radiography device designed for dental imaging. The Nomad Pro uses digital radiography to obtain images by firing an X-ray beam at an object placed in front of a digital sensor. Claws to be scanned must be placed between the incident beam and the digital sensor (Figure 3.1). The claw is positioned such that it is parallel to and in close proximity to the sensor to prevent image distortion resulting from parallax ( X-ray beams striking at an angle ≠ 90.0°) (Siu et al., 1991). Once the claw is positioned, the scanning process takes less than five minutes to perform for each specimen.

This device enabled the rapid, inexpensive, on-site acquisition of a large data set and is a practical alternative to CT scanning.

Figure 3.1: The Nomad Pro Radiography Unit. Device shown assembled and positioned for scanning. A collimated beam of X-rays is fired from the X-ray source (B) when the trigger is pulled. The protection shield (C) blocks over 99% of backscattered radiation from reaching the user(Makdissi et al., 2016). The digital sensor (A) is positioned on a block of Styrofoam directly facing the source. The sensor is 31.2 x 43.9 x 6.3 mm in dimension.

The Nomad Pro has some imaging and practical limitations that should be considered.

As the digital sensor is designed to fit inside a patient’s mouth, large claws could not be scanned using this device. The system also has limited imaging capability compared to other devices used for X-ray imaging. The intraoral digital sensor acquires images with a measured spatial resolution of 16 Lp/mm (line pairs per millimetre). This produces high quality images relative to other portable imaging devices (Pittayapat et al., 2010) but is inferior compared to devices capable of magnification radiography techniques (Boyce and Samei, 2006; Takahashi and Sakuma, 2012) or even a later model of the same device the Xios SG Supreme, which has 42

33 lp/mm theoretical resolution and 15 μm pixel size (https://www.dentsplysirona.com/en).

The equipment is sufficient for capturing details of small- to medium-sized claws but is insufficient to capture details of very small claws less than 10mm in length, and so the limited imaging capability imposes a minimum limit on claw size.

For the smallest claws in the sample with length close to 10 mm, sheath data were supplemented with photographs when necessary, and HDR (High Definition Range) Imaging

(Mann and Picard, 1994) was used to improve sharpness, particularly for the claw tip which tended to exhibit edge unsharpness (Kaestner et al., 2017). Exposure time is the only user- controlled variable for the system, and so HDR Imaging was one of few available techniques for reliably improving image quality. Multiple radiographs were taken from a fixed orientation in space at different exposure times and then layered to create a weighted average using

Helicon Focus 7 (Kozub et al., 2008) (Figure 3.2). Care was taken to ensure neither the system nor the specimen shifted between radiographs to prevent artefacts resulting from double exposure (Pongnapang, 2005).

Figure 3.2: Radiographs showing layering process. Radiographs taken with Nomad Pro. Claw is from Xiphocolaptes promeropirhynchus specimen D.1427 from the Liverpool World

Museum. (A) Single radiograph with exposure time 0.20 s. (B) Weighted average of two radiographs with exposure times 0.16s and 0.25s. (C) Weighted average of four radiographs with exposure times 0.16s, 0.19s, 0.21s, and 0.25s. (D) Weighted average of nine radiographs with exposure times 0.16s, 0.17s, 0.18s, 0.19s, 0.20s, 0.21s, 0.22s, 0.23s, and 0.25s.

Prior to operating the device, Level 1 Radiation Awareness Training course with The

University of Manchester and training under the supervision of Dr. Andrew Chamberlain were completed. To ensure safety, the device was only operated in the manner seen in Figure 3.3

(A, B) with a clearance of at least 1 m distance for any people other than the user to minimise radiation exposure. All measured dosages of radiation to the user are well below permissible limits according to the Ionising Radiation Regulations 2017 (Makdissi et al., 2016; Ionising

Radiation (Medical Exposure) Regulations 2017, 2017). In the case that museum personnel were in the same room or nearby offices, the device was not operated until the person(s) were safely behind a hypothetical plane created horizontal to the built-in shield and at least 1 m distance from the device. Warning signs were posted on doorways to the enclosed rooms in which scanning took place to notify passers-by of radiation hazard.

Figure 3.3: Positions for operating the Nomad Pro Radiography Unit (Makdissi et al., 2016).

(A) Machine held close to operator’s body and parallel to the ground. (B) Machine held at arm’s length from operator’s body and parallel to the ground. (C) Machine held with arms partially extended and perpendicular to the ground.

3.1.2 Geometric measurements

Angles of curvature for dorsal or ventral surfaces of each claw were calculated using three landmark points on the claw (Figure 3.4). Ventral and dorsal arcs of the ungual bones

(IU, OU) and claw sheaths (IS, IS2, OS) were measured according to the methods outlined in

Feduccia (1993), Pike and Maitland (2004), and Fowler and colleagues (2009). Three points

A, B, and X are placed on each claw to approximate the claw to the arc of a circle (Figure

3.4). The centre ‘O’ of this circle is drawn at the intersection of perpendicular bisectors to AX̅̅̅̅ and BX̅̅̅̅, and claw angle is taken as ∠AOB (Fenn, 2001). This measures claw arc and differs from measures of curvature that utilise the inverse of the radius (Fenn, 2001).

Figure 3.4: Methods of determining claw angle for sheaths and ungual bones. Claw drawn in

Inkscape. Measurements performed in DinoLino.exe. A_IS=base of the toepad,

A_OS=proximal dorsal end of the sheath, A_IU=base of the flexor tubercle, A_OU=proximal dorsal end of the ungual phalanx, B_U=tip of the ungual, B_S=tip of the sheath, X_IS=

¯ intersection of 퐴퐵’s bisector and the ventral curve of the keratinous sheath,

¯ X_OS=intersection of 퐴퐵’s bisector and the dorsal curve of the keratinous sheath,

¯ X_IU=intersection of 퐴퐵’s bisector and the ventral curve of the ungual phalanx,

¯ X_OU=intersection of 퐴퐵’s bisector and the dorsal curve of the ungual phalanx. (A)

Feduccia’s (1993) method of quantifying claw angle for inner curvature of the claw sheath, here denoted IS. (B) Pike and Maitland’s (2004) method of quantifying claw angle for outer curvature of the claw sheath, here denoted OS. (C) A modification of Feduccia’s method to

46 measure inner curvature of the ungual bone, here denoted IU. (D) A modification of Pike and

Maitland’s method to measure outer curvature of the ungual bone, here denoted OU.

Figure 3.5: Alternate method of measuring ventral curvature to the exclusion of the toe pad.

Claw drawn in Inkscape; measurements performed in DinoLino.exe. (A) A method of measuring claw angle after Fowler and colleagues (2009), here denoted IS2. Landmark A is located at the proximal end where the ventral surface of the keratinous sheath terminates.

Landmark B is located at the sheath tip. (B) Example Archaeopteryx claw (12th specimen, claw from right pedal D-III) exhibiting exceptional Lagerstätte preservation yet still lacking the toe pad and requiring some reconstruction (Rauhut, Foth and Tischlinger, 2018).

Two measures (IU, OU) were taken on fossilised ungual bones. For fossil claws in possession of claw sheaths, ventral curvature of the sheath was measured using a method excluding the toe pad that has also been reported to correlate with modes of life (Tinius and

Russell, 2017) (Figure 3.5). This measurement shifts landmark A distally to the start of the dorsal sheath but retains landmark B at the tip of the claw sheath as this can often be directly measured on the fossil claws (Figure 3.5). Toe pads are not known for any of the measured fossil claws, and so using this measurement as opposed to IS minimises reconstruction (Figure

3.5).

Custom software (DinoLino.exe) was created in Microsoft Visual Studio using C++ for taking measurements with improved speed as compared to alternate methods

(https://github.com/johnwelter/Dino-Lino). DinoLino.exe finds the angle of an arc of a circle using three points placed by the user on an uploaded photograph. No other software automates this procedure to such a degree, and so it is expected that removing the amount of manual input necessary should reduce measurement error.

Measurements of claw angle were taken and compared for five birds Corvus corone,

Falco rusticolus, Ixobrychus flavicollis, Dryocopus javensis, and Phasianus colchicus using

DinoLino.exe and methods used in past papers. These birds were selected for functional diversity and image quality of their radiographs. One set of measurements were taken by overlaying printed templates of concentric circles on enlarged photographs, matching a best-fit circle to the base of the claw and a point two thirds down the length, and taking claw angle of the arc after the fashion of Glen and Bennett (2007). Measurements of ventral curvature after the fashion of Feduccia (1993) and dorsal curvature after Pike and Maitland (2004) were taken manually using a goniometer and ruler and digitally using ImageJ (Abràmoff, Magalhães and

Ram, 2004). Relative variation was computed for measures taken using these methods relative to those acquired with DinoLino with the formula (A-B)/[(A+B)/2] where A is the measure taken using methodology of past authors and B is the measure taken using DinoLino (Popović and Thomas, 2017).

3.1.3 Statistical analysis

Statistical analyses and graphical summaries were performed in R (R Core Team,

2013). Box plots in the style of Tukey were created using the ggplot2 package (Wickham,

2016) for all measures of extant claws to visualise trends and outliers in the data. These initial

48 tests revealed that the measured squamate claws differ notably from avian claws, and so squamates were not included in further statistical analyses or in the final predictive model.

Variation in the data was tested using sample variance and standard deviation in

RStudio. Coefficient of variance (CV) was not used because units are the same across groups, and as CV is determined by dividing the standard deviation by the mean, this measurement is sensitive to changes in means expected to be seen among groups, particularly for ground- dwelling taxa which may have claw angle close to zero. Normality was also tested using the shapiro.test() function of the dplyr package (Wickham et al., 2015) in RStudio. The statistically significant p-value was defined as the standard 0.05.

As variances for claw angle data differ among behavioural groups (heteroscedasticity) and data for inner curvatures of the claw sheath (IS, IS2) were found to have non-normal distributions, non-parametric tests were utilised to determine if median claw angles and/or centroids of combined measures of claw angles differ by group. Permutational multivariate analysis of variance (PERMANOVA) was utilised to compare centroids for a dataset of ungual bone curvatures (IU, OU) and a dataset of all claw angles (IU, OU, IS2, OS) and the

Wilcoxon rank-sum test (also called the Mann-Whitney U test) was performed for ungual bone data. These tests do not rely on an assumption of normality and thus should produce robust results.

Body mass data were log-transformed to more closely approximate a normal distribution. The log-transformed data, original body mass data, and squared body mass data were then regressed against each measure of claw angle IU, OU, IS, OS for all extant avians to test the possibility of a linear and/or a non-linear relationship between body mass and claw angle. Simple linear regression was performed for all taxa and by group using the smatr package (Warton et al., 2012). ANOVAs were also performed between ecological group, phylogenetic order, and body mass to investigate if these variables might be influencing 49 correlations in the data. Though the Shapiro-Wilk test indicates body mass data do not have a normal distribution, ANOVAs are not very sensitive to violation of this assumption for data with moderate deviations from normality (Glass, Peckham and Sanders, 1972; Harwell et al.,

1992; Lix, Keselman and Keselman, 1996). If body mass is found to have a relationship with the data, it may be necessary to account for body size in the calculations, but attempts to remove the effects of scaling may be confounded if a complex relationship exists between qualitative factors (e.g. ecology, phylogeny), body mass, and claw angle.

Limiting the range of body masses is expected to minimise discrepancies related to unequal weight distribution among behavioural categories. The greatest weight difference between any two taxa in this study is 1894g between the Blyth’s tragopan (Tragopan blythii, mass=1930g) and the Vernal hanging parrot (Loriculus vernalis, mass=36g), and the ranges of body mass for all groups overlap. This is a much smaller disparity than the one for the sample utilised by

Birn-Jeffery and colleagues (2012), which includes ground-dwelling birds with masses up to

44,000g (Casuarius casuarius) and 23,000g (Rhea americana).

It may be useful to test the relationship between body mass and claw angle for a sample in which ecological groups have nearly identical ranges to accurately determine the relationship between claw angle and body mass. However, this would drastically reduce the number of eligible taxa, subsequently reducing sample size and phylogenetic diversity, and so this has not been done for this study.

Preliminary tests for phylogenetic signal were performed for an unrooted tree created for the sample taxa using Abouheif’s test (Abouheif, 1999) for detecting phylogenetic autocorrelation using the abouheif.moran function of the adephylo package in RStudio

(Jombart, Balloux and Dray, 2010) with 999 permutations. Abouheif’s Cmean is a version of

Moran’s I that uses a matrix of phylogenetic proximities and does not consider branch length and thus should be robust against inaccuracies in the phylogeny (Abouheif, 1999; Pavoine et 50 al., 2008). This test was reported to perform substantially better than Moran’s I and

Blomberg’s I under a Brownian motion model of evolution (Münkemüller et al., 2012).

An avian topology with nodes at order, family, and genus level was written as a .nexus file to include all extant avians in the sample and combined with the data to create phylo4d objects for performing Abouheif’s test (Jombart, Balloux and Dray, 2010). Relationships between orders and branch lengths were not specified because avian phylogenies are not well- resolved. The topology was based on the phylogeny resolved in the comprehensive study by

Jarvis and colleagues (Jarvis et al., 2014), but other recent and conflicting trees have been proposed for modern avians (Prum et al., 2015). Tests were performed to detect phylogenetic autocorrelation in datasets of each value of claw angle (IU, OU, IS, OS) and for distribution of ecological groups using binary dummy variables.

Due to the complexity of measured relationships between the phylogenetic conflict within Aves (Ericson et al., 2006; Hackett et al., 2008; Jarvis et al., 2014; Prum et al., 2015) and the even greater conflict in fossil phylogenies for the measured taxa (Xu and Zhang, 2005;

Xu et al., 2009; Lee and Worthy, 2011; Senter et al., 2012; Turner, Makovicky and Norell,

2012; Godefroit et al., 2013), phylogenetic corrective methods were not used when creating the final predictive models. It is expected that the sample is large and phylogenetically diverse enough that any results represent true ecological signal in the avian population rather than phylogenetic signal of any particular subgroup.

Measures of avian claw angle were submitted to linear discriminant analyses using the caret package in R (Kuhn, 2008). Linear discriminant analyses were performed on four subsets of the avian data: ungual bone measurements (IU, OU), claw sheath measurements

(IS, OS), sheath and bone measurements (IU, OU, IS, OS), and sheath and bone measurements with the modified measure IS2 instead of IS (IU, OU, IS2, OS). Predictive models were

51 created using the variance-covariance matrices of each LDA for each subset ungual bone curvatures, sheath curvatures, all measures, and all measures including IS2, and labelled

‘model 1’, ‘model 2’, ‘model 3’, and ‘model 4’, respectively.

Predictive success of each model was tested by classing extant birds using LDA models trained using bootstrap resampling with 2000 iterations using the caret package(Kuhn,

2008). Predictive success is gauged by accuracy at classing birds for each category, and total weighted accuracy. Success of the models is then compared to determine utility of the ungual bone measures in lieu of sheath measures and to determine to what degree is it beneficial to analyse all measures of bone and sheath curvatures.

3.2 Application of the model

Photographs of fossil specimens were acquired from published sources (Table 4.3).

The majority of fossil specimens are small-bodied paravians, avialans, and avians that were selected based on condition, image resolution, and phylogenetic closeness to extant birds.

Unfortunately, pedal claws were unavailable from scansoriopterygid specimens, but many other maniraptoran taxa with putative arboreal adaptations (e.g. Microraptor, Changyuraptor,

Pedopenna) have been included. The large-bodied, obviously ground-dwelling tyrannosaurid

T. rex and ornithomimid Beishanlong were also included to test the hypothesis that phylogenetically high angles of claw curvature influence false predictions of arboreality for predatory theropod dinosaurs (Chiappe, 1997, 2001; Padian and Chiappe, 1998).

Due to scarcity of fossil material and widespread difficulty determining factors such as age or sex of fossil animals, it was not feasible to constrain the sample of fossil taxa to only adult specimens. Some measured fossil claws may belong to immature specimens.

Ontogenetic status has been inferred for fossil specimens when possible based on descriptions

52 of histology, degree of skeletal fusion, and size in the literature, and its implications are considered in the discussion.

Taxa have been grouped according to family (Dromaeosauridae, Troodontidae,

Tyrannosauridae, Ornithomimidae, Anchiornithidae, Enantiornithidae), and clade (Avialae) based on recent descriptions. Basal avialans Archaeopteryx, Confuciusornis, Balaur, and

Sapeornis have not been grouped by family but instead have been assigned to the broad clade

“Avialae” to simplify results for the purpose of analysing broad, evolutionary trends, and to avoid ongoing disputes regarding phylogenetic status.

The selected fossil claws show no significant breakage or distortion, and so measured values are assumed to represent true claw angles of the animal during life. Slight reconstruction was necessary for all fossil sheaths and for the tips of ungual bones belonging to Balaur bondoc and two troodontid specimens Talos and MPC-D 100/140 (Figure 3.6).

Figure 3.6: Reconstructed fossil ungual bones. Reconstructions drawn using Inkscape to follow curvature of intact bone and points of termination drawn conservatively based on

53 personal observations of extant and fossil ungual bones. Pictured specimens are pedal digit III ungual bones belonging to (A) MPC-D 100/140 (B) Balaur bondoc and (C) Talos sampsoni.

Predictions were generated for all fossil taxa based on the predictive model based on ungual bone data (IU, OU) using the predict function of the stats package (R Core Team,

2013). For fossil taxa in possession of keratinous sheaths, a second set of predictions were generated using a predictive model based on an LDA of IU, OU, OS, and the modified measurement of ventral curvature of the keratinous sheath IS2. These classifications were then assessed against evidence from the literature and previous hypotheses to infer lifestyle for these taxa.

4. Results

4.1 Measurements and raw data

Full specifications of extant specimens are listed in Table 4.1. Full specifications of extinct specimens are listed in Table 4.2.

Order Binomial P S Bm g IU OU IS OS IS2 Src Spec

Predatory birds

1969- Accipitriformes Accipiter gentilis R M 912 99.6 103.6 124.5 138.2 111.8 NML 228-7 PBM Accipiter nisus R ? 237.5 89.7 97.2 128.2 111.2 82.2 unreg S Buteo buteo R ? 238 92.9 96.4 133.2 105.2 81.3 NMS unreg

Buteo buteo L ? 238 93.5 95.2 124.5 102.7 85.2 NMS unreg

MAN 1044 Circaetus gallicus R ? 1703 86.8 91.9 116.5 104.8 83.5 CH 4 Milvus migrans R M 827 84.8 86.0 151.5 125.6 115.7 NML 7960

MAN 1049 Pernis apivorus L F 832 69.5 72.7 98.5 87.5 78.1 CH 7

Cathartes aura R M 1467 82.6 86.8 85.6 96.1 78.3 NML 8360

D.296 Pandion haliaetus L ? 1489 130.0 129.6 NA NA NA NML b MAN 1056 Falconiformes Falco columbarius R ? 191 86.2 84.6 137.4 126.9 88.5 CH 1 Falco rusticolus R F 1752 99.1 102.1 139.4 117.5 98.3 NML 242b

MAN BB.61 Falco tinnunculus R F 217 89.4 92.2 138.3 118.3 105.4 CH 56 MAN 1056 Falco peregrinus R ? 782 97.5 104.0 NA NA NA CH 1 1986. Strigiformes Bubo virginianus L ? 1309 83.9 92.2 126.3 114.0 100.3 NML 2.188 MAN 1069 Strix aluco L F 524 83.9 81.4 118.3 108.0 90.1 CH 9 Strix aluco L ? 475 78.4 87.2 124.5 108.4 92.5 WC unreg

Strix aluco R ? 475 82.3 83.5 124.5 113.9 96.9 WC unreg

MAN 1070 Strix uralensis L ? 785 75.4 73.3 124.5 105.6 86.4 CH 8 PBM Tyto alba L ? 409 84.9 84.9 133.0 104.3 85.6 unreg S PBM Tyto alba R ? 409 81.1 81.3 139.8 106.6 95.7 Unreg S Tyto R ? 610 98.6 105.8 137.0 115.3 96.0 NML 285 novaehollandiae MAN Tyto tenebricosa L ? 738 94.2 99.7 133.2 123.1 99.4 B578 CH Ground-dwelling taxa

Apterygiformes Apteryx owenii L ? 1243 53.6 60.7 68.2 59.1 43.2 NML 8380

Steatornis Caprimulgiformes L ? 414 43.2 80.5 107.1 113.1 119.4 NML 1223 caripensis 1980- Charadriiformes Numenius arquata R M 742 0.0 30.5 36.0 53.1 34.8 NML 281 Larus canus L F 375 0.0 48.8 50.1 69.4 13.5 NMS 1384

Centropus 1999. Cuculiformes L ? 510 79.6 85.8 104.9 99.2 76.6 NML phasianinus 20.71 1964- Phalcoboenus Falconiformes L M 795 64.3 69.9 94.0 78.4 63.3 NML 228- megalopterus 175

Galliformes Alectoris chukar R ? 578 42.6 76.0 82.7 84.9 42.2 NML 121a

1984. Lagopus lagopus R ? 559 0.0 41.4 88.4 76.0 85.8 NML 2.197 Phasianus colchicus L ? 1133 35.4 59.3 67.6 64.7 30.2 GK unreg

Phasianus colchicus R ? 1133 37.3 57.2 61.2 65.6 33.9 GK unreg 55

Syrmaticus 1651 R M 907 29.0 52.9 55.6 66.3 46.0 NML soemmerringii 4 Tetrao tetrix R F 910 22.6 53.9 91.4 92.2 70.7 NML 1982

1984. Tragopan blythii L ? 1930 61.6 71.3 72.8 68.1 50.5 NML 2.254 Megapodius L ? 858 53.3 46.2 69.8 54.9 41.8 NML 3423a freycinet MAN Bb.71 Gruiformes Aramus guarauna L F 1080 69.3 95.5 101.2 111.6 78.4 CH 97 1984. Psophia leucoptera L ? 990 63.5 75.8 77.3 80.8 63.4 NML 2.294 Psophia viridis L F 1000 39.3 74.7 41.3 78.9 0.0 NML 253

Eupodotis Otidiformes L ? 1400 48.6 79.9 78.1 97.4 21.8 NMS unreg senegalensis Eupodotis R ? 1400 48.5 82.6 79.7 105.4 25.3 NMS unreg senegalensis Tetrax tetrax L F 750 49.8 69.2 51.0 63.9 28.2 NML 8625

Tetrax tetrax L F 750 47.7 66.1 52.7 66.4 19.1 NMS 1956

MAN Pelecaniformes Ixobrychus flavicollis L ? 318 38.4 54.2 65.7 70.5 47.8 B.602 CH Calonectris MAN Bb.84 Procellariiformes R ? 1213 54.1 66.6 110.5 81.1 54.1 diomedea CH 72

Psittaciformes Nestor notabilis R ? 868 68.1 87.9 143.9 136.5 132.7 NML 2522

1832 Strigops habroptilus R ? 1670 69.1 87.2 103.0 102.4 89.1 NML 9

Pterocliformes Pterocles orientalis R F 428 54.8 71.1 82.0 82.5 0.0 NML 578e

Crypturellus Tinamiformes R F 482 48.8 65.4 66.4 75.2 29.5 NMS 2960 obsoletus Eudromia elegans L ? 660 36.8 46.5 54.7 44.1 44.2 NML 4621

Eudromia elegans R ? 660 35.7 41.0 56.4 41.9 30.1 NML 4621

Rhynchotus L ? 900 50.7 67.4 55.9 58.9 9.8 NML 4646 rufescens Tinamus major R ? 1052 26.1 78.0 63.4 97.7 51.2 NML 3650a

1923. Tinamus major L ? 1052 32.6 75.4 57.7 86.9 46.4 NML 20.16 1867- Squamata Tiliqua scincoides L ? ? 0.0 29.7 NA NA NA NML 1 1987. Uromastyx L ? ? 46.7 39.5 46.4 52.0 0.0 NML 143.5 hardwickii 5 2000. Varanus gouldii L ? ? 41.5 49.4 69.7 70.4 29.2 NML 1.32 Perching birds 56

29-9- Bucerotiformes Tockus nasutus L M 234 62.6 93.1 118.8 118.7 115.2 NMS 1899

Columbiformes Columba oenas R F 280 47.6 93.7 111.8 125.2 88.7 NML 8083

Columba palumbus R ? 490 63.8 77.6 110.9 114.8 77.3 GK unreg

Columba palumbus L ? 490 73.7 87.4 112.2 112.7 84.8 GK unreg

1985. Ducula poliocephala R ? 537 71.7 106.7 107.3 122.6 91.1 NML 7.136 Hemiphaga 3544 L ? 600 74.0 88.7 135.0 128.2 103.0 NML novaeseelandiae d Lopholaimus R ? 518 93.9 98.0 109.1 116.6 93.8 NML 4213 antarcticus 1966. Dacelo Coraciiformes L ? 305 60.1 77.3 110.0 103.8 100.9 NML 140.1 novaeguineae 46 1920. Megaceryle maxima L ? 311 55.8 96.1 126.8 139.7 133.1 NML 146.7 Eurystomus MAN 1085 L ? 110 49.2 96.4 121.8 133.8 117.2 glaucurus CH 4

Cuculiformes Centropus menbeki R ? 510 77.3 90.7 76.5 102.3 100.4 NML 1174

MAN BD10 Cuculus canorus L ? 113 69.0 80.1 97.7 104.3 85.7 CH 281 MAN BD10 Piaya cayana L ? 108 84.3 103.3 116.6 125.0 112.4 CH 960

Galliformes Ortalis superciliaris R ? 850 60.8 75.6 89.5 97.1 64.4 NML 1644

Ortalis vetula R F 542 68.6 71.3 76.1 85.7 47.2 NML 488a

1611 Leptosomiformes Leptosomus discolor L F 220 62.8 91.2 126.4 120.0 99.2 NML 1

Musophagiformes Crinifer zonurus R M 527 73.3 93.9 123.9 115.6 103.3 NML 6265

Opisthocomus 1272 Opisthocomiformes R ? 855 88.2 96.7 119.7 120.3 99.6 NML hoazin 9 MAN Bb.70 Passeriformes Corvus corax R ? 1199 61.7 77.8 101.7 100.1 109.0 CH 43 Corvus corone R ? 570 48.3 62.1 95.4 92.7 77.8 WC unreg

Corvus corone L ? 570 53.2 63.1 108.5 101.6 84.4 WC unreg

1970. Garrulus glandarius L ? 161 66.7 81.1 107.7 102.8 97.7 NMS 76.85 0 Nucifraga MAN 1200 R ? 173 65.7 78.1 92.0 83.0 71.3 caryocatactes CH 0 MAN Bb78 Piciformes Ramphastos toco L ? 540 79.4 111.7 116.9 124.2 81.7 CH 96 13.10 Ramphastos L ? 530 77.8 100.0 128.2 122.7 116.1 NML .04.2 tucanus 2

Selenidera R F 160 88.5 101.5 117.3 115.8 103.2 NML 77a maculirostris

Psittaciformes Cacatua sulphurea L ? 340 82.6 97.1 131.1 141.1 110.3 NML T1180

Eolophus MAN R ? 320 98.2 116.6 127.0 144.0 122.3 B.68 roseicapillus CH Climbing taxa

1836 Coliiformes Colius colius L F 41.4 103.8 101.7 111.2 130.6 96.3 NML e Dendrocolaptes 8.191 Passeriformes R ? 64.2 112.3 125.3 158.5 170.3 134.1 NML certhia 0.14 Dendrocolaptes MAN R M 88.7 95.4 115.7 147.4 152.3 133.8 6196 picumnus CH Nasica longirostris L ? 85 123.0 126.0 142.3 149.8 131.3 NML 1332S

Xiphocolaptes D.14. R ? 116 86.2 112.6 112.1 129.5 103.2 NML albicollis 4b,a Xiphocolaptes D.142 L ? 155 108.5 119.9 160.7 163.7 143.1 NML major 7 Xiphocolaptes D.143 L ? 136 111.1 110.2 135.8 134.8 106.9 NML promeropirhynchus 0 Campephilus Piciformes L ? 320 98.4 138.3 151.2 152.3 124.5 NML 3777 magellanicus Campephilus L ? 511 80.7 121.2 156.0 155.2 138.6 NMS unreg principalis MAN B.438 Colaptes auratus R ? 128 96.1 106.9 123.0 134.8 120.6 CH 4 MAN BB.71 Colaptes campestris L F 165 93.6 108.7 115.0 129.7 117.8 CH 22 Dryocopus javensis L M 290 110.8 134.3 160.8 157.0 146.9 NML 3855

MAN B.598 Dryocopus martius R ? 321 98.7 129.8 135.1 157.5 135.9 CH 6 Dryocopus martius L ? 321 115.1 141.6 164.7 165.0 137.3 NML 3864

MAN Dryocopus pileatus L ? 287 112.2 125.9 153.5 156.9 141.4 7116 CH Mulleripicus R M 450 103.4 136.7 163.1 150.3 152.7 NML 4112 pulvulerulentus Mulleripicus MAN L ? 462 103.7 138.3 149.9 147.0 148.0 1978 pulvulerulentus CH MAN B.113 Picus squamatus L F 170 100.1 122.3 155.4 158.4 149.9 CH 08 Picus viridis R ? 176 94.0 120.1 148.5 133.0 133.1 NML 1324a

Megalaima MAN BD10 L ? 167 86.9 122.9 135.1 142.3 117.1 chrysopogon CH 010 MAN BD10 Psilopogon lineatus L ? 150 92.5 122.0 153.5 145.5 133.5 CH 206 25.8. Psittaciformes Ara macao R M 1015 92.3 108.4 121.6 115.4 100.9 NML 09

Loriculus vernalis L ? 36 98.9 107.9 117.9 118.8 108.3 NML 3510

Trichoglossus MAN L ? 122 85.9 106.3 112.1 116.4 73.7 unreg haematodus CH 1966. Alisterus scapularis L ? 235 79.8 97.7 121.2 123.5 112.7 NML 140.8 7 MAN Alisterus scapularis R ? 235 86.6 99.8 119.4 115.8 95.3 B.48 CH

Squamata Chamaeleo melleri R ? ? 34.3 75.8 77.7 103.8 4.0 NML 60.1

Ctenosaura R ? ? 42.2 63.1 38.7 75.3 31.3 NML 117 acanthura

Table 4.1: Sampled extant taxa. Phylogenetic orders follow a recent topology (Jarvis et al.,

2014). P=pes, S=sex, Bm g=body mass in grams, IU=inner ungual curvature, OU=outer ungual curvature, IS=inner sheath curvature (including toepad), OS=outer sheath curvature,

IS2=inner sheath curvature (not including toepad), Src=source, Spec=specimen. WC=Wild caught, GK=Game keeper. Unreg=unregistered.

Binomial Clade Pes IU OU IS OS IS2 Source Specimen Status Anchiornis Pei et al. Anch. R 77.6 63.4 NA NA NA PKUP V1068 adult huxleyi 2017 Anchiornis Anch. R 83 92.2 NA NA NA Hu et al. 2009 LPM-B00169 adult huxleyi Anchiornis Pei et al. Anch. R 71.4 53 NA NA NA BMNHC PH823 adult huxleyi 2017 Archaeopteryx Rauhut et al. Avialae R 77.9 95.5 115 98.3 57.2 12th specimen ? sp. 2018 Archaeopteryx Rauhut et al. Avialae L 88.7 81.8 115 95.5 64.3 12th specimen ? sp. 2018 Archaeopteryx MB.Av.101 (the Avialae L 84.8 97.9 123 129 89.9 MFN ? siemensii Berlin specimen) Archaeopteryx MB.Av.101 (the Avialae R 64.6 78.4 123 105 92.5 MFN ? siemensii Berlin specimen) Turner et al. Balaur bondoc Avialae L 64.6 88.1 NA NA NA EME PV.313 adult 2012 Beishanlong Makovicky et FRDC-GS Ornitho. L 56.9 60.1 NA NA NA sub-adult grandis al. 2009 JB(07)01-01 Borogovia Osmólska, Trood. R 27.8 67.8 NA NA NA ZPAL MgD-I/174 ? gracilicrus 1987 Changyuraptor Han et al. Drom. L 87 86 133 123 69.9 HG B016 adult yangi 2014 Confuciusornis Falk et al. Avialae L 87.8 94.3 153 177 151 IVPP V 13156 adult sanctus 2016

Confuciusornis Avialae R 94.5 79.3 145 134 89.3 MFN MB.Av.1158 ? sanctus Confuciusornis Wang et al. Avialae R 97.7 103 NA NA NA IVPP V 14412 A ? sanctus 2018 Confuciusornis Wang et al. Avialae L 109 111 NA NA NA IVPP V 14412 A ? sanctus 2018 Eopengornis Wang et al. Enant. L 78.6 88.4 80.3 75.7 54.5 STM24-1 sub-adult sp. 2014 Eosinopteryx Godefroit et Anch. R 54.3 63.6 NA NA NA YFGP-T5197 adult brevipenna al. 2013 Fortunguavis Wang et al. Enant. ? 71.4 110 NA NA NA IVPP V18631 adult xiaotaizicus 2014 Halszkaraptor Cau et al., Drom. ? 50.9 62.2 NA NA NA MPC D-102/109 sub-adult escuilliei 2017 Halszkaraptor Cau et al., Drom. ? 34.6 63.9 NA NA NA MPC D-102/109 sub-adult escuilliei 2017 Microraptor Hwang et al., Drom. R 79 92.1 NA NA NA CAGS 20-8-001 ? zhaoianus 2002 Microraptor Hwang et al., Drom. L 92 90.3 NA NA NA CAGS 20-8-001 ? zhaoianus 2002 Tsuihiji et al., ? Trood. L 62.9 70 NA NA NA MPC-D 100/140 ? 2016 Parapengornis Enant. L 65.3 89.6 118 109 96 Hu et al. 2015 IVPP V18632 juvenile sp. Pedopenna Sullivan et al., Anch. ? 78.7 76.7 NA NA NA IVPP V12721 ? daohugouensis 2014 Sapeornis Zheng et al. Avialae R 89.2 106 NA NA NA STM16-18 sub-adult chaoyangensis 2013 Sapeornis Avialae L 81.3 101 NA NA NA Pu et al. 2013 HGM-41HIII0405 juvenile chaoyangensis Sapeornis Avialae R 88.6 109 NA NA NA Pu et al. 2013 HGM-41HIII0405 juvenile chaoyangensis Sapeornis Gao et al. Avialae L 75.1 96.2 NA NA NA DNHM-D3078 sub-adult chaoyangensis 2012 Zanno et al. Talos sampson Trood. R 89.8 89.2 NA NA NA UMNH VP 19479 sub-adult 2011 Tyrannosaurus FMNH PR2081 Tyrann. R 44.6 85.9 NA NA NA Brochu, 2003 adult rex (“Sue”) Xiaotingia Anch. L 67.7 84.7 NA NA NA Xu et al. 2011 STM 27-2 adult zhengi Zhenyuanlong Lü & Brusatte Drom. R 76.8 88 NA NA NA JPM-0008 sub-adult suni 2015 Zhang et al. Zhouornis hani Enant. R 55.9 77.2 100 77.9 71.9 BMNHC Ph 756 sub-adult 2014 Zhang et al. Zhouornis hani Enant. L 58.6 80.3 106 84 74.7 BMNHC Ph 756 sub-adult 2014

Table 4.2: Sampled fossil taxa. IU=inner ungual curvature, OU=outer ungual curvature,

IS=inner sheath curvature (including reconstructed toepad), OS=outer sheath curvature, and

IS2=inner sheath curvature (not including toepad). Anch=Anchiornithidae,

Drom=Dromaeosauridae, Trood=Troodontidae, Tyrann=Tyrannosauridae,

Ornitho=Ornithomimidae, Enant=Enantiornithidae. 60

Table 4.3 shows variability among methods of measuring claw angle compared to the technique utilised in this study.

Best-fit circles after Glen and Bennett (2007)

Method Species IU % var OU % var IS % var OS % var

C. corone NA NA 36.0 53.20% NA NA 55.0 51.00%

D. javensis NA NA 93.0 36.30% NA NA 107.0 37.90%

Templates P. colchicus NA NA 24.0 84.80% NA NA 30.0 73.30%

F. rusticolus NA NA 63.0 47.40% NA NA 64.0 59.00%

I. flavicollis NA NA 24.0 77.20% NA NA 27.0 89.20%

Arc of a circle after Feduccia (1993) and Pike and Maitland (2004)

Method Species IU % var OU % var IS % var OS % var

C. corone 47.5 1.70% 68.0 9.10% 98.5 3.20% 98.0 5.60%

D. javensis 96.0 14.30% 129.0 4.00% 146.5 9.30% 167.5 6.50%

Manual P. colchicus 64.5 58.30% 67.5 12.90% 41.5 47.80% 62.0 4.30%

F. rusticolus 100.0 0.90% 105.0 2.80% 141.0 1.10% 120.0 2.10%

I. flavicollis 33.5 13.60% 57.5 5.90% 64.0 2.60% 64.0 9.70%

C. corone 45.9 5.10% 68.4 9.70% 103.6 8.20% 99.2 6.80%

D. javensis 111.8 0.90% 135.7 1.00% 149.3 7.40% 171.2 8.70%

ImageJ P. colchicus 50.1 34.40% 70.2 16.80% 59.5 12.70% 68.2 5.30%

F. rusticolus 105.1 5.90% 115.7 12.50% 142.7 2.30% 128.4 8.90%

I. flavicollis 44.7 15.20% 71.1 27.00% 60.4 8.40% 78.7 11.00%

C. corone 48.3 NA 62.1 NA 95.4 NA 92.7 NA

DinoLino D. javensis 110.8 NA 134.3 NA 160.8 NA 157.0 NA

P. colchicus 35.4 NA 59.3 NA 67.6 NA 64.7 NA

F. rusticolus 99.1 NA 102.1 NA 139.4 NA 117.5 NA

I. flavicollis 38.4 NA 54.2 NA 65.7 NA 70.5 NA

Table 4.3: Variability between methods of measuring claw angle. % var=percent variation from angle as determined by DinoLino. IU=inner ungual curvature, OU=outer ungual curvature, IS=inner sheath curvature (including reconstructed toepad), OS=outer sheath curvature.

Measuring claw angle using best-fit circles after Glen and Bennett ( 2007) varied the most from angles measured with the custom software. Percent variation of the claws measured using best-fit circles had up to 89.2% variation compared to measurements taken using

DinoLino. Manual measurements taken after Feduccia (1993) and Pike and Maitland (2004) differed from those taken using DinoLino by up to 58.3% for IU of P. colchicus, but most values did not vary more than 10% between DinoLino and these manual measures for most claw angles of most taxa. Digital measurements taken in the same fashion with ImageJ were the most similar to those taken by DinoLino with the maximum difference again seen for IU of P. colchicus with a 34.4% variation.

The most variation was apparent between measures of P. colchicus claw angles taken by DinoLino and other methods, followed by Ixobrychus flavicollis. These are the two flattest claws in the dataset, and the low values may be more sensitive to the calculation as they tend to yield relatively low denominators. Despite potential errors of analysis, two results are obvious. There is some difference between measures taken using the ad-hoc software

DinoLino and measures taken using past methods, and the values obtained by performing equivalent calculations in ImageJ are most similar to those obtained using DinoLino.exe.

4.2 Relationship with body mass

Shapiro-Wilk normality test data W p-value BM 0.911 2.294E-06*** IU 0.966 8.221E-03 OU 0.992 7.848E-01 IS 0.960 2.584E-03** IS2 0.966 8.389E-03** OS 0.980 1.248E-01

Table 4.4: Shapiro-Wilk normality test on claw angles and body mass. “BM” = body mass,

“IU” = Inner ungual curvatures, “OU” = outer ungual curvatures, “IS” = inner sheath curvatures, “OS” = outer sheath curvatures.

Data for body masses and ventral curvatures of the ungual bone and the claw sheath have a non-normal distribution according to Shapiro-Wilk normality tests (Table 4.4). Outer curvatures of ungual bone and claw sheath do not have a non-normal distribution according to p-values for these tests. Body mass is thus log-transformed to normalise the data for further calculations.

Figure 4.1: Histograms showing distribution of body mass and log-transformed body mass for extant avians. Histograms drawn in RStudio.

It is apparent from Figure 4.1 that the log-transformed data is less skewed than the raw data and roughly approximates a normal distribution. Claw angles unfortunately cannot be normalised in this fashion because some claw angles are equal to 0, which does not produce a real number when log-transformed.

Body mass ~ claw angle IU OU IS IS2 OS R^2 0.07 0.121 0.169 0.216 0.207 ALL p-value 0.0050** 0.0000*** 0.0000*** 0.0000*** 0.0000*** BIRDS Slope -0.0167 -0.1939 -0.0313 -0.0399 -0.0312 R^2 0.141 0.134 0.027 0 0.028 GRND p-value 0.0310* 0.0350* 0.36 0.911 0.349 Slope 0.0205 0.0152 0.0105 0.0017 0.0093 R^2 0.001 0.075 0.041 0.026 0.056 PRCH p-value 0.874 0.15 0.292 0.405 0.218 Slope -0.0016 -0.0146 -0.0146 -0.0118 -0.0143 R^2 0.081 0.1 0.12 0.003 0.004 PRED p-value 0.199 0.152 0.124 0.83 0.789 Slope 0.0069 0.008 0.008 0.0011 -0.0014 R^2 0.034 0.041 0.007 0.006 0.001 CLMB p-value 0.379 0.329 0.694 0.723 0.892 Slope -0.0099 0.0124 0.0068 0.0063 -0.0023 Body mass squared ~ claw angle IU OU IS IS2 OS R^2 0.021 0.05 0.096 0 0.122 ALL p-value 0.137 0.0190* 0.0010** 0.132 0.0000*** BIRDS Slope 0 -12.214 0 0 0 R^2 0.13 0.128 0.027 0.014 0.027 GRND p-value 0.0390* 0.0410* 0.361 0.842 0.36 Slope 0 0 0 0 0 R^2 0.003 0.071 0.044 0.008 0.068 PRCH p-value 0.773 0.163 0.274 0.636 0.173 Slope 0 0 0 0 0 R^2 0.101 0.117 0.091 0.001 0.006 PRED p-value 0.149 0.119 0.183 0.92 0.74 Slope 0 0 0 0 0 R^2 0.028 0.042 0.007 0.016 0.043 CLMB p-value 0.427 0.996 0.69 0.551 0.32 Slope 0 0 0 0 0 64

Natural log of body mass ~ claw angle IU OU IS IS2 OS R^2 0.143 0.189 0.196 0 0.243 ALL p-value 0.0000*** 0.0000*** 0.0000*** 0.241 0.0000*** BIRDS Slope -11.968 -12.214 -16.736 -20.935 -16.795 R^2 0.14 0.119 0.024 0 0.021 GRND p-value 0.0320* 0.0490* 0.388 0.959 0.419 Slope 17.21 12.024 8.351 0.6576 6.792 R^2 0 0.061 0.022 0.03 0.028 PRCH p-value 0.93 0.196 0.438 0.372 0.382 Slope -0.3496 -5.06 -3.357 -4.84 -3.915 R^2 0.042 0.058 0.127 0.019 0.003 PRED p-value 0.362 0.279 0.113 0.559 0.804 Slope 3.499 4.31 -7.403 2.095 -0.9334 R^2 0.04 0.147 0.063 0.09 0.048 CLMB p-value 0.335 0.058 0.227 0.145 0.292 Slope -2.851 6.184 5.416 6.717 4.624

Table 4.5: Linear regressions between body mass and claw angle. Statistically significant p- values are highlighted in yellow. GRND = Ground-dwelling, CLMB = Climbing, IU = Inner Ungual curvature, OU = Outer Ungual curvature, IS = Inner Sheath curvature, OS = Outer Sheath curvature

Behaviour showed a significant relationship with body mass, body mass squared, and natural log of body mass with very low p-values less than 0.05. Figure 3.1 clearly shows that medians and ranges differ among each behavioural group, and the regression between natural log of body mass and behaviour had higher R2 values (0.439) than any other variable with body mass.

Keratinous sheaths do not show a statistically significant relationship with body mass for the sample taxa. There is a positive correlation between body mass and ventral claw angle of the ungual (p=0.031) and dorsal claw angle of the ungual bone (p=0.035) and body mass for ground-dwelling taxa. However, the amount of the data explained by the linear best fit line between IU and OU angles and body masses of ground-dwelling taxa was low (R2=0.134, 65

R2=0.141), suggesting other factors are influencing this relationship. The ungual bone curvatures of ground-dwelling taxa also appeared to have a statistically significant nonlinear relationship with body mass with all p-values less than 0.05, but no R2 value exceeds 0.141, which again suggests other factors are influencing these relationships.

When all taxa were included, simple linear regressions yielded p-values much less than

0.05 for all tested relationships except IU and the square of body mass. When relationships were tested by group, only ground-dwelling taxa had p-values less than 0.05 for any regression, and no regression between sheath curvatures and body mass yielded significant p- values.

Though p-values indicate statistically significant relationship between body mass and claw angle for some regressions, two things are interesting about these results. The regressions for all taxa yield such low p-values but then when the sample is split by group, statistical significance of the relationship with body mass becomes null for all but the ungual bones of ground-dwelling taxa. This is most likely due to the great reduction of sample size from 101 claws to between 21 and 34 data points for each group, which decreases statistical power. It is also possible that relationships between behavioural group and body mass are influencing the data.

The R2 values are very low; the highest R2 value is only 0.243, and of regressions with statistically significant results, R2 values dip as low as 0.05. This means that the determined relationships, though statistically significant, are only able to explain between 5% and 24.3% of the variation. This suggests that other factors are influencing this relationship.

Figure 4.2: Box plots showing relationship between body mass and ecological group. Box plots drawn in RStudio. Shaded boxes depict interquartile range (IQR). Whiskers depict distance between IQR and points up to 1.5 distances from the IQR. Outliers are located greater than 1.5 distances from the IQR.

Figure 4.2 shows median body mass clearly differs for each ecological group. Ground- dwelling birds in the dataset have the highest body masses, followed by predatory birds, then perching birds, then climbing birds. This roughly parallels the findings of Birn-Jeffery et al.

(2012), which reported that ground-dwellers tend to be larger than birds from other ecological groups (Figure 4.2, Table 3.3), though extent of variation is restricted here by the restricted

67 range of body masses. Climbing birds have the narrowest range of body masses, followed by perching birds. Ground-dwelling and predatory birds have roughly comparable-sized ranges of body mass, though maximum, minimum, and median values differ. There is overlap between ranges for all groups.

Body mass vs ecological group ANOVA Df Sum Sq Mean Sq F value Pr(>F) Behaviour 3 6.62E+06 2.21E+06 18.13 2.89e-09 *** Tukey tests diff lwr upr p adj Ground-Climbing 6.48E+02 3.86E+02 9.10E+02 0.000 *** Perching-Climbing 1.84E+02 -7.87E+01 4.46E+02 2.65E-01 Predatory-Climbing 5.64E+02 2.69E+02 8.59E+02 1.64E-05 *** Perching-Ground -4.64E+02 -7.13E+02 2.16E+02 2.59E-05 *** Predatory--Ground -8.41E+01 -3.67E+02 1.99E+02 8.64E-01 Predatory-Perching 3.80E+02 9.76E+01 6.63E+02 3.73E-03 ** Body mass squared vs ecological group ANOVA Df Sum Sq Mean Sq F value Pr(>F) Behaviour 3 1.25E+13 4.17E+12 10.04 9.14E-06 *** Tukey tests diff lwr upr p adj Ground-Climbing 8.24E+07 3.40E+06 1.31E+07 1.41E-04 *** Perching-Climbing 1.43E+07 -3.42E+06 6.27E+06 8.67E-01 Predatory-Climbing 7.74E+07 2.30E+06 1.32E+07 1.93E-03 ** Perching-Ground -6.81E+07 -1.14E+07 -2.22E+06 1.10E-03 ** Predatory-Ground -4.99E+06 -5.72E+06 4.72E+05 9.94E-01 Predatory-Perching 6.31E+07 1.09E+06 1.15E+07 1.12E-02 * NatLog body mass vs ecological group ANOVA Df Sum Sq Mean Sq F value Pr(>F) Behaviour 3 3.33E+01 1.11E+01 25.05 7.66E-12 *** Tukey tests diff lwr upr p adj Ground-Climbing 1.5428 1.0426 2.043 0 *** Perching-Climbing 0.6809 0.1807 1.1811 3.25E-03 ** Predatory-Climbing 1.3198 0.7574 1.8821 1.00E-07 *** Perching-Ground -0.8619 -1.3359 -0.3879 4.34E-05 *** Predatory--Ground -0.2231 -0.7622 0.3161 7.01E-01 Predatory-Perching 0.6389 0.0997 1.178 1.35E-02 *

Table 4.6: ANOVAs and Tukey tests for body mass and ecological group. Calculations performed in RStudio. Post-hoc Tukey tests performed using the Tukey_HSD command after confirmation of statistical significance. Df = degrees of freedom, Sum Sq = sum of squares of deviations of all observations from the mean, Mean Sq = mean square (sum sq/df), Pr(>F) = p- value associated with the F statistic, diff = difference between means of the two groups, lwr, upr = the lower and upper end points of the 95% confidence interval, p adj = p-value after adjustment for multiple comparisons.

The results of the ANOVA provide strong evidence for linear and or nonlinear relationships between body mass and ecological groups in the dataset with very low p-values

(<0.0001) observed for the sample. Post-hoc Tukey tests for the ANOVAS between group and body mass and group and body mass squared revealed significant differences between body masses of all ecological groups except for perching-climbing and perching-ground (Table

3.3). Perching taxa are relatively intermediate with regard to body mass (Figure 4.2) and claw angles, and so this result is expected given the degree of overlap with other groups for the perching bird dataset. For the ANOVA between natural log of body mass and group, Tukey tests revealed significant differences between all but predatory-ground (Table 3.3). The statistically significant, differing results of the ANOVAs suggest there is a complex relationship between body mass and ecology.

Body mass vs order Df Sum Sq Mean Sq F value Pr(>F) Order 23 7.64E+06 3.32E+05 2.338 3.59E-03** Body mass squared vs order Df Sum Sq Mean Sq F value Pr(>F) Order 23 1.54E+13 6.68E+11 1.353 1.68E-01 NatLog body mass vs order Df Sum Sq Mean Sq F value Pr(>F) Order 23 37.11 1.6133 3.134 1.32E-04***

Table 4.7: ANOVAs for body mass and phylogenetic order. Calculations performed in

RStudio. Df = degrees of freedom, Sum Sq = sum of squares of deviations of all observations 69 from the mean, Mean Sq = mean square (sum sq/df), Pr(>F) = p-value associated with the F statistic.

The ANOVAs for phylogenetic order, body mass, and natural log of body mass have low p-values<0.05 (Table 3.4), and so the null hypothesis that body mass has no linear and/or nonlinear relationship with phylogeny is discarded. The ANOVA for body mass squared and phylogenetic order has p-value>0.05 (Table 3.4) and so these findings do not provide evidence for a relationship of this variable with phylogeny. Though there is apparently a linear and/or nonlinear relationship between phylogeny and body mass in the dataset, further tests with an expanded sample would be helpful in determining this relationship. Some orders are only represented by a single member, and so more representation within each phylogenetic group would enable more comprehensive testing of this relationship.

4.3 Relationship with phylogeny

Figure 4.3 strongly suggests that phylogenetic signal is influencing the datasets for claw angle. The calculated C-means for each claw angle place on the tail-end of the histograms of randomly calculated C-means for all measures of bone and sheath curvature IU,

OU, IS, IS2, and OS, and so there is a very low probability that these C-means arose randomly and not as a result of phylogenetic autocorrelation. P-values from the Abouheif’s test show p- values less than 0.05 for all claw angles (Table 3.5) and so the null hypothesis of no phylogenetic signal in the data is discarded.

Figure 4.3: Results of Abouheif’s test for phylogenetic autocorrelation in claw angles.

(Abouheif, 1999) Tests performed in RStudio. Histograms represent a random distribution of

999 randomised, successive squared differences between trait neighbours for 999 permutations of the phylogenetic topology. Drawn bar represents the measured C-mean for the sample. (A) C-means for inner ungual curvatures (IU) (B) C-means for outer ungual curvatures (OU) (C) C-means for inner sheath (plus toe pad) curvatures (IS) (D) C-means for inner sheath (minus toe pad) curvatures (IS2) (E) C-means for outer sheath curvatures

Test Obs Std.Obs Alter p-value IU 0.1929 2.9789 greater 0.003 OU 0.2432 3.8255 greater 0.001 IS 0.1779 2.771 greater 0.008 IS2 0.1674 2.6536 greater 0.008 OS 0.2066 3.0801 greater 0.005

Table 4.8: Results of Abouheif’s test for phylogenetic autocorrelation in claw angles.

Calculations performed in RStudio. Alter=alternate hypothesis. IU=Inner ungual bone curvature, OU=Outer ungual bone curvature, IS=Inner sheath (and toe pad) curvature,

IS2=Inner sheath (minus toe pad) curvature), OS=Outer sheath curvature.

Figure 4.4: Results of Abouheif’s test for phylogenetic autocorrelation in ecological group.

(Abouheif, 1999). Tests performed in RStudio. Histograms represent a random distribution of

999 randomised, successive squared differences between trait neighbours for 999

72 permutations of the phylogenetic topology. Drawn bar represents the measured C-mean for the sample. (A) distribution of ground-dwelling birds (B) distribution of perching birds (C) distribution of predatory birds (D) distribution of climbing birds

Test Obs Std.Obs Alter p-value GRND 0.027976 0.432242 greater 0.336 PRCH 0.086213 1.264019 greater 0.108 PRED 0.070453 1.100062 greater 0.153 CLMB 0.170335 2.688603 greater 0.007

Table 4.9: Results of Abouheif’s test for phylogenetic autocorrelation in ecological group.

Calculations performed in RStudio. Alter=alternate hypothesis, GRND=Ground-dwelling birds, PRCH=Perching birds, PRED=Predatory birds, CLMB=Climbing birds.

The results indicate that the distribution of climbing birds in the dataset is phylogenetically influenced with the C-mean placing on the tail-end of the histogram of squared differences and p-value<0.05. The null hypothesis of no phylogenetic autocorrelation in distribution of ecological category is discarded. As distribution of behavioural categories is phylogenetically influenced, it is difficult to accurately calculate and remove phylogenetic signal for claw angles without erasing true ecological signal, and so phylogenetic corrective methods are not used to create the final predictive models.

3.3 Claw geometry and behavioural category

Figure 4.5 shows the ranges of claw angle for both ungual bones (red) and sheaths

(green) for all measured taxa. The relationship found for ungual bones closely mirrors that found for claw sheaths: lower angles of curvature correlate with ground-dwelling lifestyles, intermediate claw angles correlate with perching and predatory lifestyles, and higher claw 73 angles correlate with scansorial lifestyles (Figure 4.5). Compared to claw sheaths, ungual bone curvatures had relatively constrained ranges and possessed lower values of claw angle.

Perching, ground-dwelling, and climbing taxa had roughly equivalently sized ranges of ungual bone curvatures. Predatory taxa had the smallest ranges of claw angle but, notably, were also represented by the fewest sample specimens relative to other categories. Median values of inner curvature were notably lower than those of outer curvature in all but the ‘predatory’ group.

Figure 4.5: Curvatures of D-III ungual bones and claw sheaths for all extant taxa (A)

Boxplots for inner claw curvatures (IU, IS). (B) Boxplots for outer claw curvatures (OU, OS).

Box plots created in RStudio and edited in Inkscape. Shaded boxes depict interquartile range

(IQR). Whiskers depict distance between IQR and points up to 1.5 distances from the IQR.

Outliers are represented with circles (birds) or stars (squamates) dependent on taxonomic group and are greater than 1.5 distances from the IQR.

All climbing squamates (Chamaeleo melleri, Ctenosaur acanthura) and one ground- dwelling squamate (Tiliqua scincoides) plotted as outliers and have thus been excluded from further analyses for their likely incomparability with extant avians and animals on the avian lineage. Ranges for measures taken on claw sheaths, particularly inner sheath curvatures, showed more outliers than those measured for ungual bones, suggesting greater variability of measured soft tissues. Indeed, values of claw angle measured for inner and outer sheath curvatures show greater sample variance than the same measures taken on ungual bones

(Table 3.7).

Inner Ungual Curvature (IU) GRND PRCH PRED CLMB Var 433.7075 186.1642 145.0527 123.1374 SD 20.82257 13.6442 12.04378 11.08673 Outer Ungual Curvature (OU) GRND PRCH PRED CLMB Var 248.2241 187.8561 157.6521 159.2606 SD 15.75513 13.70606 12.55596 12.61985 Inner Sheath Curvature (IS) GRND PRCH PRED CLMB Var 590.5736 225.653 205.8737 286.5927 SD 24.30172 15.02175 14.3483 16.92905 Inner Sheath Curvature (minus toepad) (IS2) GRND PRCH PRED CLMB Var 945.8823 355.8769 112.7583 307.5187 SD 30.7552 18.8647 10.61877 17.53621 Outer Sheath Curvature (OS) GRND PRCH PRED CLMB Var 445.4708 242.6626 125.045 271.9923 SD 21.10618 15.57763 11.18235 16.49219 All Claw Angles - All Taxa 75

IU OU IS IS2 OS Var 731.8036 575.5111 1034.251 1334.632 837.9557 SD 27.05187 23.98981 32.15977 36.53261 28.94746

Table 4.10: Variation in claw angle data. Var=Sample variance, SD=Standard deviation,

GRND=Ground-dwellers, PRCH=Perching birds, PRED=Predatory birds, CLMB=Climbing birds, IU=Inner ungual bone curvature, OU=Outer ungual bone curvature, IS=Inner sheath curvature (plus toepad), IS2=Inner sheath curvature (minus toepad), OS=Outer sheath curvature.

Variability differs between ecological groups, which indicates the data has a non- normal, heteroscedastic distribution. For all claw angles, data for ground-dwelling taxa showed far greater variance than did other ecological groups. Data for perching birds showed the second greatest variance regardless of measure. Data for predatory birds showed the least amount of variance for OU, IS, IS2, and OS, and had slightly greater variance than data for climbing birds in measures of inner ungual curvature (IU).

Measures of outer curvature (OU, OS) had less variance relative to measures of inner curvature (IU, IS, IS2) taken on the same structure for all taxa. Measures of sheath curvature

(IS, OS) had greater variance compared to measures of ungual bone curvature (IU, OU).

Interestingly, IS2 data had the highest variances, which contradicts previous assumptions that the high variability of inner sheath curvature results from variability of soft tissues of the toe pad (Tinius and Russell, 2017).

Though it appears that other factors are influencing variability of inner curvature such as tip wear and/or size and position of the flexor tubercle, qualitative observations of claws reveal great variation among avian toe pads, which could contribute to the high measured IS variability. The toe pads of piciform birds, for example, are small with greatly reduced fleshy

76 bits (Figure 4.6, B), the toe pads of many ground-dwelling birds are thick, long, and flat

(Figure 4.6, D), and some birds such as the kakapo have large, protruding toe pads that deform when manipulated (Figure 4.6, A).

Figure 4.6: Radiographs of avian D-III claws exhibiting significant morphological disparity.

Ungual bones outlined in yellow have inner curvatures measured at zero. (A) Left claw of kakapo (Strigops habroptila). Unregistered skin, National Museum of Scotland. (B) Left claw of ivory-billed woodpecker (Campephilus principalis). Unregistered skin, National Museum of Scotland. (C) Right claw of willow ptarmigan (Lagopus lagopus) Skin specimen

1984.2.197, Liverpool World Museum. (D) Left claw of gull (Larus canus). Skin specimen

1384, National Museum of Scotland.

Of the sampled avian ungual bones, extreme values are apparent in the ‘ground- dwelling’ category showing inner curvatures measured at zero: Numenius arquata (Eurasian curlew), Larus canus (common gull), and Lagopus lagopus (willow ptarmigan). Gulls and 77 curlews have aquatic habits and webbed feet, which may have influenced the evolution of very flat ungual bones, and the willow ptarmigan has a visibly unusual claw morphology relative to other birds (Figure 4.6). Avian claws display a wide range of morphologies and so these nearly flat ungual bones, though they plot as extreme values, are not biological outliers and represent normal diversity in the population.

Wilcoxon Rank Sum PERMANOVA Inner ungual curvature (IU) All ungual curvatures (IU, OU) CLMB GRND PRCH CLMB GRND PRCH GRND 1.07E-09 - - 0.0006 - - PRCH 1.30E-10 4.01E-06 - 0.0006 0.0006 - PRED 7.13E-03 1.25E-07 4.29E-04 0.0006 0.0006 0.0036 Outer ungual curvature (OU) All measures (IU, OU, IS2, OS) GRND 1.07E-09 - - 0.0006 - - PRCH 5.60E-08 2.84E-06 - 0.0006 0.0006 - PRED 1.36E-06 1.48E-05 1 0.0006 0.0006 0.0736

Table 4.11: Results of non-parametric tests performed on subsets of the avian data. P-values represent Bonferroni-corrected values.

Medians of inner curvatures for ungual bones (IU) differed between all behavioural categories. Medians for outer curvatures of ungual bones (OU) differed between all groups except ‘perching’ and ‘predatory’ ungual bones (Table 3.8). This is not unexpected as box plots show that claw angle ranges of the ‘predatory’ group overlap with those found for the

‘perching’ group (Figure 4.5).

Centroids generated by a PERMANOVA analysing values for both inner and outer curvatures of the ungual bone differed significantly between all groups. Interestingly, including soft tissue measures in the multivariate PERMANOVA analysis worsened

78 separation among groups. Centroids based on four measures of sheath and ungual curvatures differed significantly between all groups except for perching and predatory taxa, the comparison of which yielded a p-value equal to 0.0736 after Bonferroni correction.

Figure 4.7: LD axes based on curvatures of D-III ungual bones and claw sheaths for all extant and fossil taxa. (A) LDA plots based on LD1 and LD2 generated by an LDA of combined ungual bone measures for extant birds, overlain with data points for fossil claws (B) LDA plots based on LD1 and LD2 generated by an LDA of combined sheath and bone measures for extant birds, overlain with data points for fossil claws. Ellipses were drawn with 95% confidence from the centroid for each group. Graphs created in RStudio and edited in

Inkscape.

For the model based solely on two ungual bone measurements (Figure 4.7 A), two axes

LD1 and LD2 are generated that account for all variation. For the model utilising four measurements (Figure 4.7 B), LD1 and LD2 account for 97% of the variation, and so LD3 is not likely to impact separation. LD1 represents greater or lesser values of claw angle for any given metric, accounts for greater than 85% of variation, and separates ground-dwelling taxa from predatory/perching taxa from climbing taxa. Claws of perching birds tended to have lower LD2 values compared to claws of predatory birds, which have, on average, IU:OU roughly equal to one resulting in a slender claw with little tapering. Of the 35 fossil claws measured, 30 plotted near or above LD2=0, whereas LD2=0 represented a roughly median value for extant taxa (Figure 4.7).

Despite differing medians (Table 3.8), there is overlap between ranges for each group.

The 95% confidence ellipses created based on measures of avian ungual bones overlap between all categories except ‘ground-dwelling’ and ‘climbing’. Those based on all claw measures have similar degrees of overlap but manage to also separate ‘ground-dwelling’ and

‘predatory’ groups.

GRND PRCH CLMB PRED TOTAL Model 1 0.818 0.658 0.813 0.891 0.787 (IU,OU) Model 2 0.792 0.651 0.667 0.808 0.727 (IS,OS) Model 3 0.808 0.795 0.828 0.859 0.819 (IU,OU,IS,OS) Model 4 0.828 0.714 0.835 0.885 0.810 (IU,OU,IS2,OS)

Table 4.12: Predictive success of the models based on extant bird claws. Accuracy for each behavioural category (GRND=Ground-dwelling, PRCH=Perching, CLMB=Climbing, and

PRED=Predatory), and total weighted accuracy are listed. Model 1 utilises IU and OU, Model

2 utilises IS and OS, Model 3 utilises IU, OU, IS, and OS, and Model 4 utilises IU, OU, IS2, and OS.

For each model, average predictive success is reported for each group, and weighted average success is reported for all taxa. These metrics account for unequal sample size by category. As expected, accuracy by category varies by group and model. Of the four categories, predatory birds were consistently classed with the greatest accuracy and perching birds classed with least accuracy regardless of model. Climbing and ground-dwelling taxa were generally classed with similarly levels of accuracy by all models except the model based solely on sheath measures, which had much greater difficulty classing climbing taxa

(accuracy=0.6667) than ground-dwelling taxa (accuracy=0.7915). Perching birds were classed with lowest accuracies by all models with accuracies ranging from 0.6508 to 0.7948, which is to be expected given that perching birds possess morphologically intermediate claw angles

82 that have some degree of overlap with all other groups and are thus relatively prone to errors of misclassification (Figures 4.5, 4.7).

The predictive model based on bone and sheath measures IU, OU, IS, and OS had greatest success classing extant taxa, followed closely by the predictive model based on measures IU, OU, OS, and the modified sheath measure IS2. This is to be expected; including more variables will almost always improve predictive accuracy of the model. Of the predictive models based on only two measures, the model based on bone measures IU, OU had much greater success (total accuracy=0.7865) compared to the model based on sheath measures IS,

OS (total accuracy=0.7273). This suggests that measures taken on the ungual bone are not only be useful in lieu of claw sheaths; they may actually provide a more accurate proxy for lifestyle.

Model 4 (IU, OU, IS2, OS)

Measure LD1 LD2 LD3

Inner ungual curvature (IU) 0.039494 0.085473 -0.01741

Outer ungual curvature (OU) 0.013435 -0.11955 -0.07304

Inner sheath curvature (IS2) 0.022785 -0.00956 0.025289

Outer sheath curvature (OS) 0.001744 0.031581 0.044619

Model 1 (IU OU)

Measure LD1 LD2 NA

Inner ungual curvature (IU) 0.041889 0.082856 NA

Outer ungual curvature (OU) 0.033374 -0.09416 NA

Table 4.13: Variance-covariance loadings for each variable included in the analysis. (A)

Variance matrix based on LDA of all measures. (B) Variance matrix based on LDA of ungual bones.

4.4 Comparison with fossil taxa

For both LDA plots based either on ungual bone data or all claw measures, the majority of fossil taxa plot in the predatory/perching morphospaces (Figure 4.7 A, B).

Dromaeosaurid and anchiornithid ungual bones plot in predatory, perching, and ground- dwelling morphospaces, and avialan ungual bones plot in predatory, perching, and climbing morphospaces. Regardless of metric used, most fossil claws plot within the overlapping regions of the 95% confidence ellipses, but some taxa plot in regions distinct to a particular morphospace and can be assigned to that lifestyle with relative confidence. These include fossil dromaeosaurids Dakotaraptor and Halszkaraptor, anchiornithid Eosinopteryx, and troodontid Borogovia, which all plot as ground-dwelling, and dromaeosaurid Zhenyuanlong, which plots as perching. None of the fossil claws plot as conclusively climbing though some avialan claws including Sapeornis, Confuciusornis, and Fortunguavis plot in the shared spaces between predatory-climbing and perching-climbing taxa and could be considered candidates for scansoriality based on these results.

In the LDA plot based solely on ungual bone curvatures (Figure 4.7 A), fossil claws representing two Anchiornis and one Confuciusornis specimen plot outside the 95% confidence space. One Archaeopteryx claw lies just outside the ‘predatory’ morphospace, but when measures of the sheath are included in the analysis, this data point shifts such that it is encompassed by the ‘predatory’ morphospace. The other three Confuciusornis ungual bones

84 plotted within extant morphospaces and so the outlying Confuciusornis data point may indicate an anomalous individual rather than incomparability between modern birds and confuciusornithids altogether. However, in the LDA plot based on all claw data (Figure 4.7

B), two of the four Confuciusornis claws including the outlying specimen from the ungual- bone LDA lie outside the 95% confidence space and so it is difficult to say based on these results. Of three measured Anchiornis claws only one was recovered within the morphospaces based on extant birds, which suggests Anchiornis may have unusual claw morphologies and be incomparable to the currant sample of extant birds.

Claws belonging to Archaeopteryx, Confuciusornis, and Microraptor are morphologically very variable. Though all were classed as predatory, the four Confuciusornis claws were recovered in the overlaps between predatory, perching, and climbing morphospaces, and, in some instances, plotted outside of extant morphospace altogether.

Archaeopteryx specimens plotted within all morphospaces, and right and left digit III claws of the same specimen plotted quite far apart from each other. This was also the case for

Microraptor, for which two claws belonging to a single specimen plotted in different morphospaces (predatory and perching) and received different classifications accordingly

(Table 3.11). Conflicting predictions between feet of a single specimen also occurred for

Sapeornis and Zhouornis, but these specimens were not morphologically very variable and probably received conflicting predictions because they are recovered in areas where morphospaces overlap.

CLAW DETAILS MODEL 1 (IU + OU) MODEL 4 (IU + OU + IS2 + OS)

Binomial Specimen Pes GRND PRCH CLMB PRED GRND PRCH CLMB PRED Anchiornis PKUP R 0.061 0.092 0.001 0.846 NA NA NA NA huxleyi V1068 Anchiornis LPM- R 0.026 0.424 0.078 0.472 NA NA NA NA huxleyi B00169 85

Anchiornis BMNHC R 0.161 0.075 0.000 0.765 NA NA NA NA huxleyi PH823 Archaeopteryx 12th R 0.044 0.615 0.116 0.225 0.274 0.509 0.026 0.191 sp. specimen Archaeopteryx 12th L 0.009 0.114 0.010 0.867 0.010 0.102 0.003 0.885 sp. specimen Archaeopteryx Berlin L 0.016 0.447 0.172 0.366 0.005 0.450 0.063 0.482 sp. specimen Archaeopteryx Berlin R 0.325 0.524 0.007 0.144 0.066 0.716 0.006 0.212 sp. specimen Balaur bondoc EME PV.313 L 0.218 0.687 0.025 0.070 NA NA NA NA Beishanlong FRDC-GS L 0.715 0.184 0.101 0.000 NA NA NA NA grandis JB(07)01-01 Borogovia ZPAL MgD- R 0.952 0.048 0.000 0.000 NA NA NA NA gracilicrus I/174 Changyuraptor HG B016 L 0.014 0.199 0.024 0.764 0.005 0.125 0.002 0.868 yangi Confuciusornis IVPP L 0.011 0.311 0.098 0.580 0.000 0.205 0.046 0.750 sanctus V13156 Confuciusornis MB.Av.1158 R 0.003 0.040 0.004 0.954 0.000 0.020 0.000 0.980 sanctus Confuciusornis IVPP V R 0.002 0.168 0.271 0.560 NA NA NA NA sanctus 14412 A Confuciusornis IVPP V L 0.000 0.056 0.447 0.498 NA NA NA NA sanctus 13156 Eopengornis sp. STM24-1 L 0.047 0.460 0.045 0.448 0.404 0.359 0.016 0.221 Eosinopteryx YFGP-T5197 R 0.747 0.201 0.000 0.052 NA NA NA NA brevipenna Fortunguavis IVPP ? 0.027 0.609 0.345 0.020 0.041 0.256 0.004 0.699 xiaotaizicus V18631 Halszkaraptor MPC D- ? 0.940 0.060 0.000 0.001 NA NA NA NA escuilliei 102/109 Halszkaraptor MPC D- ? 0.817 0.155 0.000 0.028 NA NA NA NA escuilliei 102/109 Microraptor CAGS 20-8- R 0.045 0.546 0.077 0.333 NA NA NA NA zhaoianus 001 Microraptor CAGS 20-8- L 0.005 0.151 0.041 0.803 NA NA NA NA zhaoianus 001 MPC-D MPC-D 100/140 L 0.456 0.362 0.002 0.181 NA NA NA NA 100/140 Parapengornis IVPP L 0.192 0.708 0.031 0.069 NA NA NA NA sp. V18632 Pedopenna V12721 ? 0.060 0.246 0.006 0.688 NA NA NA NA daohugouensis Sapeornis STM16-18 R 0.005 0.342 0.403 0.250 NA NA NA NA chaoyangensis Sapeornis 41HIII0405 L 0.021 0.551 0.231 0.197 NA NA NA NA chaoyangensis Sapeornis 41HIII0405 R 0.004 0.320 0.525 0.151 NA NA NA NA chaoyangensis 86

Sapeornis DNHM- L 0.056 0.678 0.115 0.151 NA NA NA NA chaoyangensis D3078 UMNH VP Talos sampsoni R 0.008 0.168 0.031 0.792 NA NA NA NA 19479 Tyrannosaurus FMNH R 0.618 0.377 0.002 0.003 NA NA NA NA rex PR2081 Xiaotingia STM 27-2 L 0.202 0.633 0.019 0.147 NA NA NA NA zhengi Zhenyuanlong JPM-0008 R 0.069 0.535 0.042 0.355 NA NA NA NA suni BMNHC Ph Zhouornis hani R 0.542 0.421 0.003 0.035 0.092 0.820 0.006 0.082 756 BMNHC Ph Zhouornis hani L 0.436 0.512 0.005 0.047 0.666 0.313 0.001 0.020 756

Table 4.14: Posterior probabilities for fossil taxa generated by both predictive models. Values represent probability of each specimen belonging to an ecological grouping based on claw angles of pedal digit III. Cells highlighted in yellow represent classifications based solely on ungual bones. Cells highlighted in green represent classifications for which both this model and the model based on bone and sheath measures are in agreement. Cells highlighted in red represent classifications or which both this model and the model based on bone and sheath measures are in disagreement.

Most fossil taxa are classed as either predatory or perching. The large coelurosaurs

Tyrannosaurus and Beishanlong both received robust ground-dwelling classifications in accordance with expectations as did the basal dromaeosaurid Halszkaraptor, the troodontids

Borogovia and MPC-D 100/140, and the anchiornithid Eosinopteryx.

Of taxa for which multiple claws were measured, only three (Confuciusornis,

Anchiornis, Halszkaraptor) consistently received the same classification, though one

Anchiornis claw (LPM-B00169) was nearly classed as perching with a posterior probability equal to 0.4244 and only 0.4718 probability for a predatory classification. The model yielded conflicting predictions for claws belonging to different specimens of Sapeornis and

Archaeopteryx. There was no clear distinction between predictions for claws belonging to sub- adult specimens of Sapeornis (STM16-18 and DNHM-D3078), which received climbing and perching classifications, respectively, and for the claw of the juvenile specimen STM16-18, which received contradicting perching and climbing classifications dependent on foot.

The models occasionally yielded conflicting predictions for left and right pedal digit

III claws belonging to a single fossil specimen. This occurred for both Archaeopteryx specimens and one specimen each of Microraptor, Sapeornis, and Zhouornis. Of these individuals that received conflicting predictions, all received a perching classification for one claw, while ground-dwelling, predatory, and climbing classifications were observed for the other claw depending on the taxon.

The right Microraptor pedal D-III ungual bone was classed as perching based on a low probability (0.546) whilst the left D-III ungual bone from the same specimen was classed as predatory based on a high probability (0.843). The left ungual bone of Sapeornis specimen

41HIII0405 received a perching prediction (probability equal to 0.551) whereas the right ungual bone of the same specimen received a climbing prediction (probability equal to 0.525), but both predictions are based on low probabilities and these claws straddle the climbing- perching morphospaces (Figure 4.7). The right ungual bone of Zhouornis hani received a ground-dwelling classification based on a low probability (0.542) and the left ungual bone received a perching classification based on a similarly low probability (0.512); Figure 4.7 shows that these claws straddle the ground-dwelling and perching morphospaces. Strangely, these predictions reversed upon the inclusion of Zhouornis sheath measures, and probabilities increased for each prediction.

The right claw of the 12th specimen of Archaeopteryx sp. received perching classifications based on ungual bone and sheath measures, but its left claw received predatory classifications from both models. The Berlin specimen of Archaeopteryx also received 88 conflicting predictions dependent upon claw; strangely, the left claw received conflicting perching and predatory predictions depending on whether sheath measures were included

(Table 3.11). Only one of three Archaeopteryx claws was classified as predatory by the model based on ungual bones, and though two were classified as predatory by the model based on all measures; the claw that shifted from a perching to a predatory classification has a nearly equivalent probability of perching (0.45) as it does for predatory (0.482) when classed using model 4.

Roughly half of the specimens that were classified using both predictive models received conflicting predictions dependent on model. These shifts tended to occur for specimens classed with relatively low posterior probabilities that essentially straddle two morphospaces; Zhouornis claws, for example, receive roughly 50% perching and 50% ground-dwelling probabilities from model 1 and receive more polarised classifications from model 4. The posterior probabilities for the left claw of the Berlin specimen of Archaeopteryx were relatively intermediate between perching (0.4504) and predatory (0.4822) in the model including all measures (Figure 4.7 B). Eopengornis martini shifts from a perching classification to a ground-dwelling classification with 59.6% probability. Parapengornis eurycaudatus shifts from a perching classification to a predatory classification with 61.7% probability.

Though most taxa that received different predictions only required a slight change in posterior probabilities to shift, two taxa (Parapengornis, Eopengornis) received drastically different posterior probabilities upon the inclusion of sheath measures. Parapengornis shifts from a perching classification with .7075 probability to a predatory classification with 0.6992 probability. Eopengornis shifts from having a 0.4599 probability of perching lifestyle and only

0.0469 probability of a ground-dwelling lifestyle to receive a ground-dwelling classification with 0.4042 posterior probability in the second model. Though Eopengornis still may be 89 considered as having an intermediate morphology with these posterior probabilities and based on its location in the LDA plots, this is still quite a large change. 5. Discussion

5.1 Findings on claw ecomorphology

This study finds a similar relationship between lifestyle and curvature for ungual bones as has been found for claw sheaths in which claw arc directly increases with arboreality and predatory habits (Pike and Maitland, 2004; Glen and Bennett, 2007; Birn-Jeffery et al., 2012), but ungual bones possess low claw angles compared to sheaths (Figure 4.5) and thus cannot be directly compared to extant sheaths without risking misclassification.

The dorsal arcs of the ungual bones of climbing squamates had higher measured curvatures compared to ground-dwelling squamates in the sample (Figure 4.5), which could suggest that the results of this study, to some extent, exhibit a proposed universal trend among tetrapod claws (i.e. increasing claw curvature with increasing arboreality). However, such a relationship was less clear for measurements of the ungual bone’s ventral curve, and the measured squamates possessed much lower claw angles compared to extant birds and thus may be incomparable using these methods (Figure 4.5). These results contradict a study finding squamate claws to be generally congruent with avian claws across behavioural categories (Birn-Jeffery et al., 2012). The apparent incongruence between avian and squamate ungual bones in this study may stem from the very small sample size (n=5), or perhaps ungual bones are affected more strongly than claw sheaths by phylogenetic/biomechanical differences between clades.

The predictive model based on ungual bone curvatures outperformed the one based on curvatures of the keratinous sheath (Table 3.9), which indicates that ungual bones provide a more accurate metric. Reconstructing external claw morphology is thus unnecessary for

90 comparative analysis. However, predictions based only on fossil ungual bones shifted with the inclusion of sheath measures for nearly half of the measured taxa (Table 3.11). In most instances, these shifts occurred for taxa with intermediate claw morphologies for which only slight changes were necessary to yield a different classification.

Previous studies have reported conflicting results regarding which, if any, of the defined groups exhibits greater behavioural generalisation with regard to claw shape (Pike and

Maitland, 2004; Birn-Jeffery et al., 2012). The results of this study suggest that claw sheaths may more greatly reflect behavioural generalisation or specialisation, whereas ungual bones appear to possess roughly equivalent spread by group (Figure 4.5). Ungual bone curvatures of the predatory sample taxa have the narrowest morphospace, and personal observations suggest most raptorial claws have a similar narrow shape with moderate to strong curvature. A specialised form for predatory talons may relate to biomechanics of piercing and/or grip during prey capture/dispatching, or perhaps the predatory taxa could be interpreted as hyper- specialised perching birds with claws that must be adapted to many requirements (e.g. grasping branches, grasping struggling prey, piercing flesh) and thus cannot deviate far from a certain functional morphology. It is also plausible that the narrow ‘predatory’ morphospace is an artefact of the relatively small sample size for predatory claws (n=22).

Dorsal curvatures of ungual bones and claw sheaths were found to have higher values of claw angle (Figure 4.5), generally less variability compared to ventral curvatures taken on the same structure (Table 3.7), and generally similar relationships with ecology, phylogeny, and body mass (Tables 4.4, 3.4, and 3.7; Figures 4.2, 4.3, 4.4, and 5.1). Dorsal claw angles were also found to have a normal distribution, whereas ventral claw angles had a non-normal distribution (Table 4.3) and are thus less likely to yield accurate results when analysed using parametric tests.

The measure IS, which includes the toe pad, was unexpectedly less variable than the similar measure IS2, which does not include the toe pad. No measure of sheath curvature (IS,

IS2, OS) exhibits a statistically significant relationship with body mass by group (Table 4.4), which is ideal for this and similar studies, but regressions performed between body mass and sheath curvatures for all birds were significant and so it is unclear to what extent scaling impacts curvature of avian claw sheaths (Table 4.4). IS appears to be slightly more useful as a predictor of avian lifestyle based on the predictive success of model 3 (accuracy=0.819) versus model 4 (0.8103), but the difference is not great.

5.2 Potential sources of error

5.2.1 Effects of scaling on form

Figure 5.1 clearly shows that very little of the variation in claw angle can be explained by a linear relationship with body mass although there is a weak association with inner and outer curvatures of the ungual bones for ground-dwelling taxa. There is a notable amount of scatter not encompassed by the best-fit lines or the 95% confidence interval. The amount of scatter is here interpreted to indicate an absence of strong linear scaling signal in the data, but it is also possible body mass estimates are contributing to noise as the data points represent generic estimates of specific weight rather than known masses for individual specimens.

Figure 5.1: Regression plots for claw angle against body mass. Scatterplots created in

RStudio. Edits done in Inkscape. The shaded area represents 95% confidence limits of the

93 best-fit line for the regressions. Only two of the regressions are statistically significant when performed by group. (A) Linear model for (Inner ungual curvature ~ ln(Body mass)) (B)

Linear model for (Outer ungual curvature ~ ln(Body mass)) (C) Linear model for (Inner sheath curvature + toepad ~ ln(Body mass)) (D) Linear model for (Inner sheath curvature - toepad ~ ln(Body mass)) (E) Linear model for (Outer sheath curvature ~ ln(Body mass))

It is also apparent from figure 5.1 that the relationship between body mass and claw angle differs by ecological group. Slopes of the best-fit lines visibly differ, with ground- dwelling birds consistently showing a positive correlation between claw angle and body mass

(i.e. curvature increases with body size) whereas perching birds tended to have a negative correlation between claw angle and body mass (i.e. curvature decreases with scaling). The relation for predatory birds and climbing birds was more complex and differed dependent on measure. For predatory birds, curvatures of the ungual bone and IS2 increase as does ln(body mass), whereas other measures of the sheath (IS, OS) decrease with ln(body mass). For climbing birds, all measures of curvature showed a positive correlation with ln(body mass) except inner curvature of the ungual bone (IU), which shows a negative correlation with ln(body mass).

Regressions between body mass and claw angle for all extant birds recovered significant relationships between body mass and dorsal/ventral curvatures for ungual bones and sheaths (Table 4.5). However, best-fit lines cannot explain much of the variation

(0.000 ≤ R2 ≤ 0.243), which suggests that other factors are influencing the recovered relationships and parallels findings of past studies (Birn-Jeffery et al., 2012; Tinius & Russell

2017). Confounding factors probably include phylogeny and ecology, both of which display significant relationships with body mass (Tables 4.5 and 4.6, Figure 5.1) and claw angle

(Tables 4.7, 4.8, 4.9; Figures 4.2, 4.3, and 4.4). It is also possible that uneven distribution of 94 body mass across ecological groups (Figure 4.2, Table 4.6) is influencing the measured relationships, but this effect should be greatly reduced compared to past studies that do not restrict body mass for extant taxa (Birn-Jeffery et al., 2012; Pike & Maitland, 2004; Tinius and

Russell, 2017). It is unclear whether the relationship between body mass and claw angle is linear or nonlinear as significant relationships are recovered between claw curvatures and body mass, body mass squared, and log-transformed values of body mass.

Regressions performed by group revealed that the relationship between claw angle and body mass differs between ecological groups (Figures 4.2 and 5.1, Table 4.4 and 4.5) as has been reported by some past studies (Pike and Maitland, 2004; Birn-Jeffery et al., 2012). Only the ungual bones of ground-dwelling birds exhibit a statistically significant relationship with body mass (Table 4.4); this contradicts a previous study recovering significant relationships between claw angle and scaling for all groups (Pike and Maitland, 2004). Reduced significance for regressions performed by group may stem from reduced sample sizes; however, Birn-Jeffery and colleagues (2012) also reported only some behavioural groups

(ground-dwelling, predatory) display a significant relationship between claw angle and body mass for a much larger sample, which suggests that the relationships recovered by this study represent real trends in the population. The complexity of the relationship between scaling and claw angle hinders statistical correction, but as the best-fit lines explain very little of the variation and scaling appears only to affect the claws of ground-dwelling taxa, body mass may not present a large source of error.

5.2.2 Phylogenetic influence on form

It has been suggested that there is a phylogenetic component to avian behaviours and ecology (Irwin, 1996; Böhning-Gaese and Oberrath, 1999), and it is apparent upon reviewing the literature that climbing and predatory birds have relatively limited phylogenetic 95 distribution compared to ground-dwelling and perching birds (Hoyo, Elliott and Sargatal,

2001, 2002, 2004, 1992, 1996, 1999; Hoyo et al., 1994, 1997; Hoyo, Elliott and Christie,

2003, 2005, 2006, 2008; Hoyo, Christie and Birdlife, 2007). All climbing birds that have scansorial habits into adulthood and raptors that seize prey using the talons belong to three to four orders within Telluraves (Yuri et al., 2013; Prum et al., 2015), whereas perching birds and ground-dwelling birds are represented by >10 orders distributed across each major clade

(Table 4.1). The Abouheif’s tests recovered significant phylogenetic autocorrelation in distribution of climbing birds, which suggests that phylogenetic distribution of behaviour may influence the data (Figure 4.3, Table 3.5). However, this may stem from over-representation of piciform birds in the dataset (Table 4.1) and could potentially be resolved by sampling a more diverse group of climbing taxa.

Unfortunately, the Abouheif’s tests recovered significant phylogenetic autocorrelation for all measures of claw angle (Figure 4.4, Table 3.4). Utilising phylogenetic generalised least squares (pGLS) for the regressions and/or phylogenetic ANOVAs to test the effects of scaling and the relationship between claw angle and ecological group (Table 4.11) would have improved robustness of these results. If the qualitative variables (phylogeny, ecology) were unrelated, then phylogenetic corrective methods could be employed to create a more robust predictive model. However, ecology appears to have a phylogenetic distribution (Figure 4.3,

Table 3.5), and so employing methods to correct for phylogenetic signal in the quantitative variables is likely to erase true ecological signal.

Though phylogenetic signal contributes a potential source of error and lessens the robustness of results, the sample is phylogenetically diverse, and the predictive models performed with high accuracies when classifying extant taxa (Table 3.9) and yield predictions consistent with other features for many fossil taxa. It thus seems unlikely that the influence of phylogenetic signal negates results of the model. 96

The set of robust ground-dwelling predictions for a diverse group of non-avian theropods indicates that the model is capable of distinguishing between ground-dwelling and arboreal theropods despite claims that phylogenetically high claw curvatures of non-avian theropods influence false hypotheses of arboreality (Chiappe, 1997, 2001; Padian and

Chiappe, 1998). T. rex and Beishanlong receive ground-dwelling classifications, which is to be expected based on their large sizes (>150kg) and cursorial adaptations (Naish, 2000b). The dromaeosaurid Halszkaraptor, the troodontid Borogovia, and the anchiornithid Eosinopteryx also received ground-dwelling classifications (Table 3.11) consistent with osteological features (Osmólska, 1987; Cau et al., 2017) and integumentary structures (Godefroit et al.,

2013).

5.2.3 Overlapping ranges: form and behaviour

Though median claw angles were statistically significant between most behavioural categories, median claw angles could not always be separated between perching and predatory birds (Table 3.8) and there is significant overlap between the 95% confidence ellipses drawn for each behavioural category, particularly perching birds (Figure 4.7). This most likely causes the frequent misclassifications of perching birds relative to other groups (Table 3.9). This is concerning given the prevalence of perching classifications for the fossil taxa (Table 3.11).

Predictions of lifestyle for fossil taxa based on claw morphology alone (or in fact any behavioural/ecological inference) should be considered alongside additional evidence to improve reliability of predictions (Lauder and Thomason, 1995; Hone and Faulkes, 2014).

Behavioural complexity presents an issue for this and any study attempting to link morphology with mode of life (Glen and Bennett, 2007; Birn-Jeffery et al., 2012). Most animals utilise pedal claws for multiple functions to varying degrees, and so it is difficult to class any animal into a single behavioural category (Glen and Bennett, 2007; Birn-Jeffery et 97 al., 2012). Many birds with perching or climbing adaptations also spend time foraging on the ground, and all predatory birds measured for this study are also perching birds. One could alternately interpret the ‘predatory’ morphospace as a ‘predatory-perching’ morphospace and the perching morphospace as ‘non-predatory perching’.

Unfortunately it was not feasible to include ground-dwelling taxa that utilise claws for prey capture/dispatching (e.g. secretary birds) as their claws were too large, and so this data cannot be used to test if there is a distinction between claw morphologies of ground-dwelling and arboreal predators. More research on predatory claws would be useful in resolving why

Talos, which has an apparently ground-dwelling lifestyle, is grouped in the same category as taxa with apparent arboreal adaptations (e.g. Changyuraptor, Pedopenna).

Peters and Görgner (1992) have suggested that claw curvature cannot reliably be correlated with an arboreal lifestyle because cliff-dwelling birds (e.g puffins) also possess sharply-curved claws to cling to rocks. It is true that this hinders classification for a fossil taxon like Archaeopteryx, for which habitat structure is contested due to all specimens being allochthonous (Chiappe, 1997; Yalden, 1997; Burnham, 2007). However, no cliff habitats are known for Solnhofen (Naish, 2000b), and for the autochthonous fossils from the forests of

Jehol (Zhou, Barrett and Hilton, 2003) (Microraptor, Changyuraptor, Xiaotingia,

Zhenyuanlong, Sapeornis, Confuciusornis, Fortunguavis, Eopengornis, Parapengornis) or

Daohugou (Yongdong et al., 2006) (Pedopenna), a cliff-dwelling ecology is insufficient to explain high pedal claw curvatures seen in these taxa. It seems reasonable to conclude that the curved claws of taxa living in forested palaeoenvironments represent arboreal adaptations.

Cliff-dwelling taxa are not represented in the dataset, and so further work including cliff- dwelling taxa may be useful to test if the claws of these taxa may be distinguished from those of arboreal birds.

5.2.4 Conflicting predictions

Inspection of the fossil material revealed that some individuals that received conflicting predictions dependent on foot possess claws in variable preservational conditions

(Figure 5.2). The right ungual bone of Microraptor (CAGS 20-08-0001) is broken near the tip

(Figure 5.2 A); this could have altered claw angle from its in-vivo value. The left claw sheath of the 12th specimen of Archaeopteryx is more degraded than the right claw sheath (Figure 5.2

C, D), which could have influenced imprecise landmark selection. The right claw of the Berlin specimen appears bent near the tip (Figure 5.2 E), but the effect is slight and it is unclear whether this results from taphonomic distortion or natural variation. The claws of these individuals (Archaeopteryx, Microraptor) plot relatively far apart from each other in the LDA plots (Figure 4.7 A, B), and so it seems most likely for these taxa that conflicting predictions result from distortion or degradation in the fossil material.

Figure 5.2: Right and left pedal D-III claws of specimens that received conflicting predictions. (A) Right claw of Microraptor specimen CAGS 20-8-001 (B) Left claw of

100

Microraptor specimen CAGS 20-8-001 (C) Right claw of Archaeopteryx 12th specimen (D)

Left claw of Archaeopteryx 12th specimen (E) Right claw of Archaeopteryx Berlin specimen

(F) Left claw of Archaeopteryx Berlin specimen (G) Right claw of Sapeornis specimen

41HIII0405 (H) Left claw of Sapeornis specimen 41HIII0405 (I) Right claw of Zhouornis specimen BMNH Ph 756 (J) Left claw of Zhouornis specimen BMNH Ph 756.

The claws of Zhouornis and the juvenile Sapeornis specimen (41HIII0405) also received conflicting predictions dependent on foot. Image quality is poor for the left claw of

Sapeornis (Figure 5.2 H) as this was unfortunately the only available image in the literature, and so landmark locations are less clear than they are for the right claw. Both claws of

Zhouornis are well-preserved and pictured in high-resolution, but the flexor tubercle has a strange, sloping morphology in the left claw (Figure 5.2 J), which may indicate post-mortem distortion or an abnormal morphology/pathology, and so it was difficult to locate landmark

A_IU. Though unclear landmarks likely contributed some error, measured angles did not differ greatly between left and right Sapeornis (41HIII0405) or Zhouornis claws (Table 4.3).

Zhouornis and Sapeornis claws have intermediate morphologies that plot in the overlap between morphospaces (Figure 4.7) and received classifications based on low posterior probabilities (Table 3.11), and so it seems likely that conflicting predictions for these individuals occurred due to slight, naturally-occurring differences between feet.

Conflicting predictions also occurred between specimens for Sapeornis, and

Archaeopteryx. These could occur due to intraspecific variation, which has been reported to significantly impact ungual bone form for extant taxa (MacLeod and Rose, 1993). Bone strengthens in response to loadings (Wolff, 1986), and so any variation in lifestyle, even between individuals of the same species, could influence different claw angles for the ungual

101 bone. For taxa with intermediate morphologies such as Sapeornis, the claws of which plot relatively close together but recover in the overlap between morphospaces (Figure 4.7), this could influence conflicting classifications. However, Archaeopteryx claws are morphologically variable (Table 4.3, Figure 4.7) and intraspecific variation is unlikely to explain the degree of variation in claw angle observed for these taxa.

The Archaeopteryx specimens have been interpreted as different species by some workers (O’Connor et al., 2013; Lefèvre et al., 2014; Foth and Rauhut, 2017), and so it is possible that these conflicting predictions occur because specimens represent different species with different modes of life. It is also possible that some of the fossil specimens represent different growth stages with different lifestyles compared to adults of the same species. The

Sapeornis specimens represent sub-adult and juvenile growth stages (Gao et al., 2012; Pu et al., 2013; Zheng et al., 2013), and ontogenetic status is unclear for the Archaeopteryx specimens (Houck, Gauthier and Strauss, 1990; Bennett, 2008; Erickson et al., 2009).

5.2.5 Incomparable fossil morphologies and behaviours

Pedal claws of some coelurosaurian taxa (Halszkaraptor, T. rex, Xiaotingia) plot outside the ‘predatory’ morphospace (Figure 4.7 A) and receive low posterior probabilities for a predatory classification (Table 3.11) despite these taxa possessing clear adaptations to a predatory lifestyle. The ‘ground-dwelling’ predictions generated for Halszkaraptor and T. rex claws may not preclude a predatory function; the functional demands of locomotion constitute a strong selective pressure, and so morphology is likely to reflect an animal’s primary mode of locomotion even for multifunctional pedal claws. Alternately, these classifications may occur because these taxa preferentially utilized manus claws and/or jaws rather than pedal claws for predatory functions as has been suggested based on biomechanical analyses (Carpenter, 2002).

Manual claws of non-avian theropods are likely to have more obvious adaptations to 102 performing ecological functions as they would have been free during terrestrial locomotion.

Unfortunately, this study is limited to comparing homologous structures, and extant birds, with the exception of some juvenile taxa (e.g. hoatzins) (Hoyo, Elliott and Sargatal, 1996) do not have functional manual claws.

Anchiornis claws provide a good example of an incomparable fossil morphology. All

Anchiornis claws received a predatory classification, and two of the three specimens received very high probabilities for a predatory classification (0.8458, 0.7657). However, the claws that receive high probabilities of predatory lifestyle plot well outside extant morphospaces (Figure

4.7). The model utilises high LD2 values to distinguish the ‘predatory’ morphospace, and so the model classes the Anchiornis claws as predatory because they possess very high values for

LD2. However, the values of LD2 are so high for Anchiornis claws that they are incomparable to claws of extant birds. Consequently, not much can be concluded regarding ecology/lifestyle of Anchiornis using these methods.

5.3 Predictions for fossil taxa

The results suggest scansorial habits for many of the measured fossil taxa with many grouping within or close to the perching morphospace with roughly two-thirds plotting outside the 95% confidence ellipse for ground-dwelling taxa (Figure 4.7).

Some taxa that receive perching classifications (Balaur, Microraptor, Xiaotingia, and arguably Archaeopteryx) or plot near/within the perching morphospace and possess putative arboreal adaptations (Changyuraptor, Pedopenna) lack an opposable hallux and would have been incapable of perching in the style observed in modern birds. However, primitive birds and/or non-avian theropods may have been able to perch in different manners. Hallux reversal is not a binary state; some theropod taxa possess a distally situated (e.g. Microraptor) or

103 medially directed (e.g. Archaeopteryx) hallux that could have enabled rudimentary perching and grasping of branches (Mayr and Peters, 2007; Hattori, 2016). Additionally, many extant taxa that do not possess reversed halluces on the pes are able to rest comfortably on tree branches (e.g. tree kangaroos, sloths, some felids).

Some workers have speculated that non-avian theropods and basal avialans lack certain characters (e.g. prehensile tails, reversible ankles) that have convergently evolved in many lineages of arboreal marsupials, eutherians, and squamates (Naish, 2000a; Dececchi and

Larsson, 2011), and so no coelurosaur could have been more than facultatively arboreal.

However, many extant climbing animals (e.g. hyraxes, grey foxes, goats, and some lizards) do not exhibit climbing specialisations (Naish, 2000a) and claw-based climbers tend to be less arboreally adapted (Dececchi and Larsson, 2011). It is also possible that different developmental paths between taxa lead to the evolution of divergent forms to perform similar functions. For example, it has been suggested that the stiffened tail of dromaeosaurs could have been used as a prop for support on tree trunks in the fashion of woodpeckers (Chatterjee,

1997; Chatterjee and Templin, 2004).

Reversible ankles enable many extant arboreal taxa to descend head-first down tree trunks, but for taxa like Microraptor or Archaeopteryx that probably were capable of gliding

(Chatterjee and Templin, 2003, 2007; Alexander et al., 2010), head-first climbing descent would have been unnecessary. Even birds with compromised wings and juveniles that do not yet have flight feathers are able to control their descent by flapping (Dial, Randall and Dial,

2006), and so theropods with long feathers on the arms may have been able to jump/glide down from elevated perches and flap their “proto-wings” to control descent (Dial, 2003).

Additionally, small body sizes would have lessened impact upon falls from tree branches for some taxa (Chatterjee, 2015), and so it seems feasible that any small, feathered theropods could have safely descended from a tree by simply jumping from their perch. 104

5.3.1 Archaeopteryx

Archaeopteryx claws were classed as either perching or predatory, and three of the four measured received very low posterior probabilities for a ground-dwelling classification (Table

3.11). One could interpret Archaeopteryx as a perching bird based on classifications for well- preserved claws and relatively high posterior probabilities. Predatory classifications seem less likely to be correct based on relatively poor condition of claws receiving this classification

(Figure 5.2), but a predatory lifestyle has been suggested for Archaeopteryx based on features including sharp, conical teeth and sharp claws on the manus that may have been used to seize prey (Ostrom, 1974; Paul, 1988). However, archaeopterygid claws have also been interpreted as having a climbing function based on curvatures and geometry (Yalden, 1985, 1997;

Feduccia, 1993; Burnham et al., 2011), and it seems that claw morphology cannot provide support in favour of either hypothesis. Archaeopteryx could more specifically be interpreted as having a raptorial, perching lifestyle with pedal claws adapted to predatory and perching functions. A ground-dwelling, predatory interpretation does not explain numerous apparently arboreal adaptations seen in Archaeopteryx.

Archaeopteryx probably lived in or near the islands of Solnofen, where there is evidence of trees up to 3m tall in which the animal could have perched (Barthel, Swinburne and Morris, 1990). An arboreal/perching lifestyle is supported by phalangeal proportions, wrist flexibility, hallucal orientation, and apparent flight adaptations (e.g. pennaceous feathers, a robust furcula, and anatomy of the shoulder girdle). The elongation of penultimate phalanges in the pes and manus (Mayr and Peters, 2007) indicate grasping ability which, in conjunction with its swivel wrist joints, could have facilitated climbing of tree trunks in the fashion of juvenile hoatzins (Chatterjee, 1997; Hopson, 2001; Fowler, Freedman and Scannella, 2009), and the medially directed hallux indicates rudimentary but not specialized perching ability 105

(Yalden, 1997; Mayr and Peters, 2007; Hattori, 2016). An arboreal glider model could explain why Archaeopteryx possessed flight adaptations (e.g. pennaceous feathers, a robust furcula, a specialized shoulder girdle, hindlimb feathers) but did not possess musculoskeletal characters necessary for launching into flight from a run (Chatterjee and Templin, 2003; Longrich, 2006;

Burnham, 2007). An animal adapted to gliding, presumably among tree branches, but incapable of launching into flight from the ground must have a method of scaling a tree and, to some extent, grasping and locomoting amongst branches.

The results are in conflict between perching and predatory but lean toward an arboreal interpretation for Archaeopteryx based on classifications for better preserved pedal claws

(Figure 5.2), low posterior probabilities for a ground-dwelling classification, and high probabilities of perching for all but the poorly preserved left claw of the 12th specimen (Table

3.11). Unfortunately this study cannot offer much clarification as Archaeopteryx claws were well-scattered, recovered within shared morphospaces for perching, predatory, ground- dwelling, and climbing taxa (Figure 4.7) and received conflicting predictions between feet for both specimens (Table 3.11).

5.3.2 Maniraptoran dinosaurs

The majority of measured maniraptoran claws (9 of 14) received perching and/or predatory classifications, are characterised by low posterior probabilities for ground-dwelling classifications, and plot near or within the 95% confidence ellipse for perching taxa (Table

3.11, Figure 4.7). These results could be interpreted as supporting scansoriality in maniraptoran dinosaurs such as Microraptor, Changyuraptor, Xiaotingia, and Pedopenna that possess adaptations consistent with an arboreal lifestyle (Xu and Zhang, 2005; Alexander et al., 2010; Han et al., 2014). Four maniraptoran specimens (Halszkaraptor, Borogovia,

Eosinopteryx, and specimen MPC-D 100/140) also received ground-dwelling classifications 106 consistent with skeletal structures and proportions, histology, integumentary structures, and size (Table 3.11; Figure 4.7) (Osmólska, 1987; Godefroit et al., 2013; Tsuihiji et al., 2016;

Cau et al., 2017). These results indicate behavioural diversity in maniraptoran dinosaurs and shed light on evolutionary trends.

It has been suggested based on pedal proportions, feather arrangement, and cranial capacities that over time, dromaeosaurids moved away from an ancestral cursorial state and became arboreally adapted, whilst troodontids became better adapted to cursorial locomotion

(Chatterjee and Templin, 2004; Fowler et al., 2011). The results appear to mirror this proposed trend; terrestriality/arboreality as measured along LD1 roughly correlates with phylogenetic status of measured deinonychosaurs as determined by a recent phylogeny (Shen et al., 2017) (Figure 4.7). It is unclear to what extent claw curvatures correlate with cursorial or arboreal ability as this study only investigated correlations with broad behavioural categories, but the results display an interesting trend that could merit future research.

The basal-most troodontid measured (Talos) plots with raptorial birds, the more derived, unnamed specimen MPC-D 100/140 plots closer to the ground-dwelling morphospace, and the most derived troodontid measured (Borogovia) plots firmly within the ground-dwelling morphospace (Shen et al., 2017) (Figure 4.7). Though locomotory implications of a predatory classification are unclear, Talos is most likely a ground-dwelling predator as it has very short forelimbs inconsistent with arboreality (Zanno et al., 2011). A specialized predatory function for the pedal claws of Talos is supported by pathological modification of the second digit on the left pes of this specimen consistent with high loadings related to predatory/defensive functions (Zanno et al., 2011). The ground-dwelling classification (probability equal to 0.456) is difficult to assess for the more derived troodontid specimen MPC-D 100/140 due to the fragmentary nature of the specimen, but proportions of preserved metatarsals and hind limb elements suggest that specimen MPC-D 100/140 was 107 more cursorially adapted than basal forms (Fowler et al., 2011; Tsuihiji et al., 2016). The ungual bone from the most derived troodontid measured, Borogovia, received a strong ground-dwelling classification (probability equal to 0.952), and fossil evidence suggests

Borogovia was well adapted to cursorial locomotion based on its long, slender metatarsals and hindlimbs (Osmólska, 1987; Fowler et al., 2011).

Conversely, the basal-most dromaeosaur measured (Halszkaraptor) plots with ground- dwelling taxa, and multiple small, derived dromaeosaurs fall within (Microraptor,

Changyuraptor) or close to (Zhenyuanlong) the overlap between ‘predatory’ and ‘perching’ morphospaces (Figure 4.7), which suggests that the foot may have been utilised for both prey capture/immobilization (Fowler et al., 2011) and grasping branches in a niche analogous to that of modern raptorial birds.

The strong ground-dwelling classifications for both Halszkaraptor ungual bones

(probabilities equal to 0.847, 0.940) are consistent with anatomical features and histology

(Cau et al., 2017). Halszkaraptor may represent an abnormal data point as it has been interpreted as a semi-aquatic animal and a sub-adult based on histology (Cau et al., 2017).

However, semi-aquatic birds clustered with ground-dwelling taxa, and so it is expected that the halszkaraptorine claw should not possess unusual adaptations or be incompatible with these methods.

Though it received a predatory classification (probability equal to 0.764),

Changyuraptor recovers within the shared morphospace between perching and predatory taxa

(Figure 4.7) and seems a likely candidate for arboreality based on various anatomical features consistent with scansoriality including relatively long forelimbs, small size, and phalangeal proportions, hindlimb feathering, pennaceous feathers suggestive of gliding ability, and a distally located hallux (Han et al., 2014). Depending on how the predatory group is interpreted here, this classifications may or may not lend support to arboreal hypotheses. 108

Though it received a perching classification, Zhenyuanlong seems less likely to have had arboreal habits as it has relatively short arms, is large compared to other “winged” dinosaurs (165 cm estimated length) (Lü and Brusatte, 2015), and is classified as perching based on a low posterior probability (0.535). A perching ecology is not impossible for this taxon; some large-bodied birds (e.g. turkeys) are capable of perching (Hoyo et al., 1994), and

Zhenyuanlong has long, flight-adapted feathers and sharply curved claws. Though the animal was probably too large to fly or glide, it may have utilized its developed proto-wings to flap during WAIR for scaling tree trunks (Dial, 2003; Lü and Brusatte, 2015).

Though Microraptor claws received both perching and predatory predictions (Table

3.11), it is not unreasonable to conclude that sharply curved claws in taxa such as Microraptor were adapted for an arboreal, raptorial lifestyle. This is supported by a specimen preserving an articulated enantiornithine bird in the gut (O’Connor, Zhou and Xu, 2011) (Figure 5.3), biomechanical models indicating gliding ability (Chatterjee and Templin, 2007; Alexander et al., 2010), and certain anatomical features including hindlimb feathers that would have hampered terrestrial locomotion (O’Connor, Zhou and Xu, 2011).

109

Figure 5.3: Specimen of Microraptor gui (IVPP V17972) with remains of enantiornithine bird preserved in abdominal cavity and close-up of line drawing of IVPP V17972.

Fem=femor, uln=ulna, rad=radius, hum=humerus, tmt=tarsometatarsal, thv=thoracic vertebrae(O’Connor, Zhou and Xu, 2011).

The gut contents in specimen IVPP V1792 (Figure 5.3) are perhaps the most compelling evidence for arboreal habits in a non-avian theropod and strongly support our predatory-perching interpretation of Microraptor. The enantiornithine remains overlay the right ribs and are overlain by the left ribs and gastralia and so may be confidently interpreted as gut contents rather than an associated fossil. The articulated nature of the enantiornithine

110 remains suggests it was hunted rather than scavenged. Enantiornithines are widely accepted to have been arboreal for numerous adaptations seen in the skeleton (O’Connor, Zhou and Xu,

2011; O’Connor, 2019), and so evidence of active predation upon an enantiornithine bird by

Microraptor gui suggests that Microraptor must have been capable of arboreal locomotion in order to catch a tree-dwelling bird under normal circumstances. The model of Microraptor arboreality is somewhat weakened by the discovery of fish remains and a mammal jaw in other specimens, which would suggest the animal was not strictly arboreal but rather an opportunistic predator (Xing et al., 2013).

Anchiornithids (Anchiornis, Eosinopteryx, Xiaotingia, Pedopenna) received a wide range of predictions (predatory, perching, ground-dwelling). These results could be interpreted as supporting behavioural diversity within the clade. Alternately, the great variability between these taxa could indicate that the recent phylogeny grouping these taxa in their own family

Anchiornithidae is inaccurate (Foth and Rauhut, 2017). The phylogenetic status of these

“anchiornithids” is the subject of much dispute, and it is plausible that these taxa could be less closely related than Foth and Rauhut (2017) reported (Xu and Zhang, 2005; Xu et al., 2009;

Lee and Worthy, 2011; Senter et al., 2012; Turner, Makovicky and Norell, 2012; Godefroit et al., 2013). Phylogenetic dissimilarity between the taxa could provide an explanation for why anchiornithid claw morphologies are so diverse (Figure 4.7).

The anchiornithid Eosinopteryx received a robust ground-dwelling classification

(Table 3.11) consistent with anatomical features and reduced plumage on the hindlimbs

(Godefroit et al., 2013). Conversely, the anchiornithids Xiaotingia and Pedopenna seem likely candidates for arboreality based on and various anatomical features consistent with scansoriality including relatively long forelimbs, small size, and phalangeal proportions (Xu and Zhang, 2005; Xu et al., 2011). Pedopenna also possesses extensive hindlimb feathering and pennaceous feathers suggestive of gliding ability (Xu and Zhang, 2005). Xiaotingia 111 receives a perching classification, but Pedopenna recovers within the overlap between predatory and perching morphospaces and is characterized by a predatory classifications with

0.688 probability. Depending on how the predatory group is interpreted, this classification may or may not lend support to an arboreal interpretation for Pedopenna. These taxa were discovered in Liaoning from the Daohugou (Pedopenna) and Jehol formations (Xiaotingia), both of which had forested palaeoenvironments within which scansoriality would have been an adaptive trait.

5.3.3 Avians and avialans

Claws from Sapeornis were classed as either perching or climbing, either of which could be argued for the taxa based on skeletal characters included a reversed hallux and extremely elongated forelimbs (Zhou and Zhang, 2002). No Sapeornis claw measured in this study belongs to a mature adult, and claws received conflicting predictions, and so further research would be useful when interpreting sapeornithid ecology.

The perching classification for the strange, island-dwelling taxon Balaur bondoc, here interpreted as an avialan, is supported by some fossil evidence including small size, pedal anatomy suited for grasping, and enlarged, hyperextensible claws on pedal digits I and II that could have functioned as “climbing crampons” (Manning et al., 2005; Fowler, Freedman and

Scannella, 2009; Csiki et al., 2010). Its hindlimb proportions have been interpreted as having a stabilising function during locomotion over uneven ground; perhaps a stabilising function of the hindlimbs instead (or additionally) enabled locomotion on the complex surfaces of the arboreal canopy (Dececchi and Larsson, 2011; Brusatte et al., 2013). The hand is fused and atrophied, which indicates poor grasping ability, but its exposed, curved manual claws may have still functioned during arboreal locomotion, and its forelimbs are relatively long

112 compared to the hindlimbs, which indicates they would have been utilised during climbing and/or grasping of branches (Brusatte et al., 2013).

Interestingly, all claws belonging to Confuciusornis were classed as predatory by all models (Figure 4.7, Table 3.11). This is unexpected given that the fossil evidence strongly supports arboreality for Confuciusornis; claw and toe proportions as well as the presence of an opposable hallux indicate specialised perching ability (Zhou and Farlow, 2001), and it has been suggested that confuciusornithids were specialist trunk-climbing birds based on elongated forelimbs and phalangeal proportions (Zhang et al., 2009). Three of four

Confuciusornis claws recover in the overlapping morphospace between predatory, climbing, and perching taxa, which could be interpreted as supporting an arboreal lifestyle (Figure 4.7); however, the locomotory implications of a ‘predatory’ classification are unclear and not much can be said about confuciusornithid arboreality based on these results.

A predatory interpretation is controversial but not entirely unsupported; it has been suggested that softness of the horny beak is a taphonomic artefact (O’Connor, 2019), some aspects of cranial morphology have been interpreted as evidence that Confuciusornis was a

“sally-striking predator” that snatched insects on the wing (Elzanowski, Peters and Mayr,

2018), and fish remains have, though somewhat equivocally, been labelled as gut contents for one specimen (Dalsätt et al., 2006). It should be noted that a predatory lifestyle for

Confuciusornis seems unlikely even given the above-mentioned evidence. Being insectivorous differs from being predatory as defined by this study as an insectivorous bird does not need talons adapted to catching prey. Conclusive direct evidence of diet is unknown for

Confuciusornis specimens (O’Connor, 2019), but morphology and inferred softness of the beak are traditionally regarded as evidence of herbivory in Confuciusornis (Hou et al., 1999;

Zinoviev, 2009). As some data points for Confuciusornis plot outside of extant morphospaces

113

(Figure 4.7), and carnivory is not well-supported by the literature, more evidence would be needed to assert a predatory lifestyle for Confuciusornis with confidence.

Arboreality is often assumed based on the presence of a reversed hallux because this character enables true perching, but hallux reversal is the ancestral state for birds and might only signify phylogenetic status for some taxa. All extant ground-dwelling birds also possess reversed halluces, and so further evidence is useful when inferring behaviours for primitive birds.

Enantiornithine birds Zhouornis, Eopengornis, Parapengornis, and Fortunguavis received predictions of ground-dwelling and perching ecologies but generally could not be well-resolved. Early birds were well-diversified by the late Mesozoic (Bell and Chiappe,

2011; O’Connor, 2019), and so it is possible that ground-dwelling and perching lifestyles are represented in the dataset. However, Zhouornis received conflicting predictions for claws of the same specimen, and the perching prediction for Fortunguavis contradicts interpretations of a scansorial lifestyle based on skeletal characters (Wang, O’Connor and Zhou, 2014). The

LDA plots show Fortunguavis plots on the edge of the climbing morphospace (Figure 4.7 A) and receives relatively high probability of climbing (0.3451), and so the claw angles observed for this taxa are low for a climbing bird but not so low that they preclude the possibility of a scansorial ecology.

Bohaiornithids are generally interpreted as arboreal birds based on high claw curvatures (M. Wang et al., 2014), but the bohaiornithid Zhouornis hani measured in this study has relatively flat claws that yielded high posterior probabilities for a ground-dwelling lifestyle. The Zhouornis specimen (BMNHC Ph 756) is interpreted as a sub-adult based on degree of skeletal fusion (Zhang et al., 2014) and it is possible that ontogenetic status is contributing to this unexpected result. It is also possible that previous interpretations of bohaiornithid arboreality do not accurately represent diversity within the clade. 114

Pengornithids have been interpreted as specialized vertical climbers based on a short, robust, grasping foot, an expanded pygostyle, and potentially stiffened tail feathers (Hu,

O’Connor and Zhou, 2015). Each character interpreted as arboreal for its similarity to those of modern woodpeckers and the Brown Creeper (Certhia americana) (Hu, O’Connor and Zhou,

2015). However, both pengornithids (Parapengornis, Eopengornis) in the dataset were classified as perching birds by the model (Table 3.11) and plotted outside the 95% confidence ellipse for extant climbing birds (Figure 4.7 A). The Eopengornis specimen (STM24-1) is interpreted as a sub-adult based on histology and degree of skeletal fusion (Wang, O’Connor and Zhou, 2014), and the Parapengornis specimen (IVPP V18632) is interpreted as a juvenile or very young sub-adult based on histology (Hu, O’Connor and Zhou, 2015). The probable immature status of both pengornithids in the dataset could be influencing morphology in unexpected ways (X. Wang et al., 2014).

6. Conclusions

If we are to take these results at face value then it paints an interesting picture of the evolution of arboreality in the avian lineage. In conjunction with other lines of evidence, the results suggest that at least some of the fossil paravians had arboreal habits. The prevalence of

‘perching’ classifications for fossil paravians and early avialans suggests that arboreal habits evolved in some ancestor of birds prior to the origination of Aves. Though true perching in the style of extant birds may have been impossible for non-avian theropods, it is plausible that curved pedal claws evolved to grip branches and/or tree trunks in conjunction with manual claws to enable arboreal locomotion.

The results suggest scansoriality rather than strict arboreality for several non-avian theropods including Archaeopteryx, Microraptor, Xiaotingia, and Balaur. A specialist trunk-

115 climbing ecology could be inferred based on these results and fossil evidence for the avialan

Sapeornis and possibly the basal avians Fortunguavis and Confuciusornis, all of which had ungual bones that recovered within the 95% confidence ellipse for climbing taxa and possess adaptations consistent with climbing ability. Unless claw phenotypes differed significantly between extant avians and Mesozoic theropods, it seems likely that any arboreal habits in

Mesozoic theropods would have been facultative or rudimentary. Perhaps some small, feathered theropods scaled trees on occasion to escape predators and/or chase prey, had a clumsy gait amongst the branches, and descended by gliding or jumping down as suggested by

Dial (2003).

Unfortunately, many fossil claws in the sample possess intermediate morphologies that are recovered in the overlap between morphospaces. It is important to view posterior probabilities and LDA plots in addition to classifications to understand the strength of any given prediction. Viewing the LDA plots in addition to posterior probabilities also enables one to verify that fossil claws have comparable morphologies to extant claws (Figure 4.7, Table

3.11).

Conflicting predictions occurred between feet of individuals and between individuals of the same genera in some instances (Table 3.11; Figure 5.2). Conflicting predictions could be caused by natural variation, preservational condition (e.g. taphonomic distortion), ontogenetic and/or taxonomic differences between individuals, abnormal morphologies, inaccurate identification of landmarks, or some unknown factor. Regardless of causative factors, conflicting predictions hinder classification of fossil taxa and should be considered as a source of error in this study.

The study found that curvature of the pedal digit III ungual bone not only provides a useful proxy for certain modes of life but in fact exhibits a stronger correlation with these lifestyles (accuracy=0.7865) than do similar measures taken on claw sheaths 116

(accuracy=0.7273). By determining the relationship between ungual bone curvature and lifestyle for a phylogenetically diverse sample of extant birds, this study overcomes limitations of past studies and introduces a more reliable way to compare fossil and extant claws. In addition, analysis of fossil ungual bones minimises amount of necessary reconstruction. Analysing ungual bone morphology thus has clear benefits in palaeontological study over the analysis of external claw morphology.

The results indicate dorsal and ventral claw curvatures provide different information and so the inclusion of both inner and outer curvatures is likely to improve predictive strength of the model. If one plans to utilise a single metric of claw curvature, lower variabilities and normal distributions of dorsal curvatures should be considered against the additional information provided by ventral curvatures, which have more specialised forms for interacting with the substrate (Tinius and Russell, 2017).

Though this study focused on ungual bones, interesting results were also found regarding claw sheath curvatures. Including the toe pad in measurements unexpectedly increases variability in the data, and so any useful information gained from including this structure is arguably negated by heightened variability and the potential error introduced by reconstructing fleshy toe pads for fossil claws. One could justify usage of either IS or IS2 based on variability, preservation status, and research enquiry.

Though this analysis minimizes error related to reconstruction of fossil material and recovers robust relationships for extant birds (Tables 3.7 and 3.8), the results show that reconstructing fossil behaviour from D-III claw curvature is not particularly reliable (Table

3.11). Comparative analyses of ungual bones may still be vulnerable to confounding influences of phylogenetic signal, effects of scaling, behavioural complexity, overlapping morphospaces, taphonomic distortion of fossil material, and incomparability between extant and fossil taxa. 117

Utilising solely curvatures of the D-III ungual bone to predict lifestyle is ill-advised because morphospaces overlap, and it is difficult to determine if fossil and extant claws are truly comparable based on these results. This would be more useful as part of a more detailed morphological analysis measuring multiple skeletal characters, and further work is needed to resolve hypotheses for fossil taxa. Curvatures of fossil ungual bones may be useful when studying fossil taxa as they exhibit a strong correlation with behaviour and ecology for modern birds and squamates, but other lines of evidence such as palaeoenvironmental data, evidence of diet, biomechanics, and other relevant features such as size and feathering should also be considered.

118

References

Abouheif, E. (1999) ‘A method for testing the assumption of phylogenetic independence in comparative data’, Evolutionary Ecology Research. 1(8), 895–909. Abràmoff, M. D., Magalhães, P. J. and Ram, S. J. (2004) ‘Image processing with ImageJ’, Biophotonics international. 11(7), 36–42. Agnolín, F. L. and Novas, F. E. (2013) ‘Systematic Palaeontology’, in Agnolín, F. L. (ed.), and Novas, F. E. (ed.) Avian Ancestors. Springer, pp. 9–36. Alexander, D. E., Gong E., Martin L. D., Burnham D. A., and Falk A. R. (2010) ‘Model tests of gliding with different hindwing configurations in the four-winged dromaeosaurid Microraptor gui’, Proceedings of the National Academy of Sciences. 107(7), 2972– 2976. Alexander, R. M. (1968) ‘Animal mechanics’. University of Washington Press. Alexander, R. M. (1989) ‘Optimization and gaits in the locomotion of vertebrates’, Physiological reviews. 69(4), 1199–1227. Alexander, R. M. (2003) Principles of animal locomotion. Princeton University Press. Alfaro, M. E., Bolnick, D. I. and Wainwright, P. C. (2005) ‘Evolutionary consequences of many-to-one mapping of jaw morphology to mechanics in labrid fishes’, The American Naturalist. 165(6), E140–E154. Alibardi, L. (2008) ‘Microscopic analysis of lizard claw morphogenesis and hypothesis on its evolution’, Acta Zoologica. 89(2), 169–178. Alibardi, L. (2009) ‘Development, comparative morphology and cornification of reptilian claws in relation to claws evolution in tetrapods’, Contributions to Zoology. 78(1), 25– 42. Bakker, R. T. (1972) ‘Anatomical and ecological evidence of endothermy in dinosaurs’, Nature. 238(5359), 81. Bakker, R. T. (1986) The dinosaur heresies. William Morrow. Barthel, K. W., Swinburne, N. H. M. and Morris, S. C. (1990) Solnhofen. CUP Archive. Bell, A. and Chiappe, L. M. (2011) ‘Statistical approach for inferring ecology of Mesozoic birds’, Journal of Systematic Palaeontology. 9(1), 119–133. Bennett, S. C. (2008) ‘Ontogeny and Archaeopteryx’, Journal of Vertebrate Paleontology. 28(2), 535–542. Biewener, A. A. (1983) ‘Locomotory stresses in the limb bones of two small mammals: the ground squirrel and chipmunk’, Journal of Experimental Biology. 103(1), 131–154. Biewener, A. and Patek, S. (2018) Animal locomotion. Oxford University Press. Birn-Jeffery, A. V., Miller C. E., Naish D., Rayfield E. J., and Hone D. W. E. (2012) ‘Pedal claw curvature in birds, lizards and Mesozoic dinosaurs–complicated categories and compensating for mass-specific and phylogenetic control’, PLoS One. 7(12), e50555. Bock, W. (1983) ‘On extended wings’, The Sciences. 23(2), 16–20. Bock, W. J. (1965) ‘The role of adaptive mechanisms in the origin of higher levels of organization’, Systematic Zoology. 14(4), 272–287. Böhning-Gaese, K. and Oberrath, R. (1999) ‘Phylogenetic effects on morphological, life- history, behavioural and ecological traits of birds’, Evolutionary Ecology Research. 1(3), 347–364. Boyce, S. J. and Samei, E. (2006) ‘Imaging properties of digital magnification radiography’, Medical physics. 33(4), 984–996. Brusatte, S. L., Vremir M., Csiki-Sava Z., Turner A. H., Watanabe A., Erickson G. M., and Norell M. A. (2013) ‘The osteology of Balaur bondoc, an island-dwelling 119

dromaeosaurid (Dinosauria: Theropoda) from the Late Cretaceous of Romania’, Bulletin of the American Museum of Natural History. 2013(374), 1–100. Burgers, P. and Chiappe, L. M. (1999) ‘The wing of Archaeopteryx as a primary thrust generator’, Nature. 399(6731), 60. Burnham, D. A. (2007) ‘Archaeopteryx–a re-evaluation suggesting an arboreal habitat and an intermediate stage in trees down origin of flight’, Neues Jahrbuch für Geologie und Paläontologie-Abhandlungen. 245(1), 33–44. Burnham, D. A., Feduccia A., Martin L. D., and Falk A. R. (2011) ‘Tree climbing–a fundamental avian adaptation’, Journal of Systematic Palaeontology. 9(1), 103–107. Carpenter, K. (2002) ‘Forelimb biomechanics of nonavian theropod dinosaurs in predation’, Senckenbergiana lethaea. 82(1), 59–75. Carrano, M. T. (2006) ‘Body-size evolution in the Dinosauria’, Amniote paleobiology: perspectives on the evolution of mammals, birds, and reptiles. 8, 225–268. Cartmill, M. (1974) ‘Pads and claws in arboreal locomotion’, in Jenkins Jr., F. A. (ed.) Primate locomotion. Elsevier Inc. pp. 45–83. Cartmill, M. (1985) ‘Climbing’, in Hildebrand, M. (ed.), Bramble, D. M. (ed.), Liem, K. F. (ed.), and Wake, D. B. (ed.) Functional vertebrate morphology. Harvard Univ. Press. pp. 73-88. Cau, A., Beyrand V., Voeten D. F. A. E., Fernandez V., Tafforeau P., Stein K., Barsbold R., Tsogtbaatar K., Currie P. J., and Godefroit P. (2017) ‘Synchrotron scanning reveals amphibious ecomorphology in a new clade of bird-like dinosaurs’, Nature. 552(7685), 395. Chatterjee, S. (1997) ‘The Rise of Birds. 312 pp’. The Johns Hopkins University Press, Baltimore. Chatterjee, S. (2015) The rise of birds: 225 million years of evolution. JHU Press. Chatterjee, S. and Templin, R. J. (2003) ‘The flight of Archaeopteryx’, Naturwissenschaften. 90(1), 27–32. Chatterjee, S. and Templin, R. J. (2004) ‘12. Feathered Coelurosaurs from China: New Light on the Arboreal Origin of Avian Flight’, in Currie, P. J. (ed.), Koppelhus, E. B. (ed.), Shugar, M. A. (ed.), and Wright, J. L. (ed.) Feathered Dragons: Studies on the Transition from Dinosaurs to Birds. Indiana University Press, p. 251. Chatterjee, S. and Templin, R. J. (2007) ‘Biplane wing planform and flight performance of the feathered dinosaur Microraptor gui’, Proceedings of the National Academy of Sciences. 104(5), 1576–1580. Chiappe, L. M. (1997) ‘Climbing Archaeopteryx’, A response to Yalden. Archaeopteryx, 15, 109–112. Chiappe, L. M. (2001) ‘Mesozoic birds’, First Edition. e LS. Wiley Online Library. Crandell, K. E., Herrel A., Sasa M., Losos J. B., and Autumn K. (2014) ‘Stick or grip? Co- evolution of adhesive toepads and claws in Anolis lizards’, Zoology. 117(6), 363–369. Csermely, D. and Rossi, O. (2006) ‘Bird claws and bird of prey talons: Where is the difference?’, Italian Journal of Zoology. 73(01), 43–53. Csermely, D., Rossi, O. and Nasi, F. (2012) ‘Comparison of claw geometrical characteristics among birds of prey and non-raptorial birds’, Italian Journal of Zoology. 79(3), 410– 433. Csiki, Z., Vremir M., Brusatte S. L., and Norell M. A. (2010) ‘An aberrant island-dwelling theropod dinosaur from the Late Cretaceous of Romania’, Proceedings of the National Academy of Sciences. 107(35), 15357–15361. Czerkas, S. A. and Yuan, C. (2002) ‘An arboreal maniraptoran from northeast China’, The Dinosaur Museum Journal, 1, 63–95. 120

Dalsätt, J., Zhou Z., Zhang F., and Ericson P. G. P. (2006) ‘Food remains in Confuciusornis sanctus suggest a fish diet’, Naturwissenschaften. 93(9), 444. Davies, L. (1987) ‘Long adult life, low reproduction and competition in two sub‐Antarctic carabid beetles’, Ecological Entomology. 12(2), 149–162. Dececchi, T. A. and Larsson, H. C. E. (2011) ‘Assessing arboreal adaptations of bird antecedents: testing the ecological setting of the origin of the avian flight stroke’, PLoS One. 6(8), e22292. Dial, K. P. (2003) ‘Wing-assisted incline running and the evolution of flight’, Science. 299(5605), 402–404. Dial, K. P., Jackson, B. E. and Segre, P. (2008) ‘A fundamental avian wing-stroke provides a new perspective on the evolution of flight’, Nature. 451(7181), 985. Dial, K. P., Randall, R. J. and Dial, T. R. (2006) ‘What use is half a wing in the ecology and evolution of birds?’, BioScience. 56(5), 437–445. Dunning Jr, J. B. (2009) CRC handbook of avian body masses, 2nd edition. CRC press. Elzanowski, A., Peters, D. S. and Mayr, G. (2018) ‘Cranial morphology of the Early Cretaceous bird Confuciusornis’, Journal of Vertebrate Paleontology. 38(2), e1439832. Erickson, G. M. Anderson C. L., Britton T., Elzanowski A., Johansson U. S., Källersjö M., Ohlson J. I., Parsons T. J., Zuccon D., and Mayr G. (2009) ‘Was dinosaurian physiology inherited by birds? Reconciling slow growth in Archaeopteryx’, PLoS One. 4(10), e7390. Ericson, P. G. P., Anderson C. L., Britton T., Elzanowski A., Johansson U. S., Källersjö M., Ohlson J. I., Parsons T. J., Zuccon D., and Mayr G. (2006) ‘Diversification of Neoaves: integration of molecular sequence data and fossils’, Biology letters. 2(4), 543–547. Fain, M. G. and Houde, P. (2004) ‘Parallel radiations in the primary clades of birds’, Evolution. 58(11), 2558–2573. Feduccia, A. (1993) ‘Evidence from claw geometry indicating arboreal habits of Archaeopteryx’, Science. 259(5096), 790–793. Feduccia, A. (1999) The origin and evolution of birds. Yale University Press. Fenn, R. (2001) Geometry. London: Springer. Foth, C. and Rauhut, O. W. M. (2017) ‘Re-evaluation of the Haarlem Archaeopteryx and the radiation of maniraptoran theropod dinosaurs’, BMC evolutionary biology. 17(1), 236. Fowler, D. W., Freedman E. A., and Scanella J. B. (2011) ‘The predatory ecology of Deinonychus and the origin of flapping in birds’, PLoS One. 6(12), e28964. Fowler, D. W., Freedman, E. A. and Scannella, J. B. (2009) ‘Predatory functional morphology in raptors: interdigital variation in talon size is related to prey restraint and immobilisation technique’, PloS one. 4(11), e7999. Fukunaga, K. (2013) Introduction to statistical pattern recognition. Elsevier. Gao, C., Chiappe L. M., Zhang F., Pomeroy D. L., Shen C., Chinsamy A., and Walsh M. O. (2012) ‘A subadult specimen of the Early Cretaceous bird Sapeornis chaoyangensis and a taxonomic reassessment of sapeornithids’, Journal of Vertebrate Paleontology. 32(5), 1103–1112. Gere, J. M. and Timoshenko, S. (1985) ‘Mechanics of materials’, Stanford University, PWS Publishers, Massachusetts, USA, Printed by Wadsworth International, California, USA, ISBN: 0-534-05196-0. Glass, G. V., Peckham, P. D., and Sanders, J. R. (1972) ‘Consequences of failure to meet assumptions underlying the fixed effects analyses of variance and covariance’, Review of educational research. 42(3), 237-288. 121

Glen, C. L. and Bennett, M. B. (2007) ‘Foraging modes of Mesozoic birds and non-avian theropods’, Current biology. 17(21), R911–R912. Godefroit, P., Demuynck H., Dyke G., Hu D., Escuillié F., and Claeys P. (2013) ‘Reduced plumage and flight ability of a new Jurassic paravian theropod from China’, Nature Communications. 4, 1394. Greenwold, M. J., Bao W., Jarvis E. D., Hu H., Li C., Gilbert M. T. P., Zhang G., and Sawyer R. H. (2014) ‘Dynamic evolution of the alpha (α) and beta (β) keratins has accompanied integument diversification and the adaptation of birds into novel lifestyles’, BMC evolutionary biology. 14(1), 249. Hackett, S. J., Kimball R. T., Reddy S., Bowie R. C. K., Braun E. L., Braun M. J., Chojnowski J. L., Cox W. A., Han K-L., and Harshman J. (2008) ‘A phylogenomic study of birds reveals their evolutionary history’, Science. 320(5884), 1763–1768. Hamrick, M. W. (2001) ‘Development and evolution of the mammalian limb: adaptive diversification of nails, hooves, and claws’, Evolution & Development. 3(5), 355–363. Hamrick, M. W. (2003) ‘Evolution and development of mammalian limb integumentary structures’, Journal of Experimental Zoology Part B: Molecular and Developmental Evolution. 298(1), 152–163. Han, G., Chiappe L. M., Ji S-A., Habib M., Turner A. H., Chinsamy A., Liu X., and Han L. (2014) ‘A new raptorial dinosaur with exceptionally long feathering provides insights into dromaeosaurid flight performance’, Nature Communications. 5, 4382. Harwell, M. R., Rubinstein E. N., Hayes W. S., Olds C. C. (1992) ‘Summarizing Monte Carlo results in methodological research: The one-and two-factor fixed effects ANOVA cases’, Journal of educational statistics. 17(4), 315-339. Hattori, S. (2016) ‘Evolution of the hallux in non-avian theropod dinosaurs’, Journal of Vertebrate Paleontology. 36(4), e1116995. Hawkins, D. M. (2004) ‘The problem of overfitting’, Journal of chemical information and computer sciences. 44(1), 1–12. Hone, D. W. E. and Faulkes, C. G. (2014) ‘A proposed framework for establishing and evaluating hypotheses about the behaviour of extinct organisms’, Journal of Zoology. 292(4), 260–267. Hopson, J. A. (2001) ‘Ecomorphology of avian and nonavian theropod phalangeal proportions: implications for the arboreal versus terrestrial origin of bird flight’, New perspectives on the origin and early evolution of birds. 211–235. Hou, L., Martin L. D., Zhou Z., Feduccia A., and Zhang F. (1999) ‘A diapsid skull in a new species of the primitive bird Confuciusornis’, Nature. 399(6737), 679. Houck, M. A., Gauthier, J. A. and Strauss, R. E. (1990) ‘Allometric scaling in the Earliest Archaeopteryx lithographica’, Science. 247(4939), 195–198. Hoyo, J. del, Elliot A., and Christie D. A. (1994) Handbook of the birds of the world. Volume 2: New world vultures to guineafowl. Barcelona: Lynx Edicions. Hoyo, J. del, Elliot A., and Christie D. A. (1997) Handbook of the birds of the world. Volume 4: Sandgrouse to cuckoos. Barcelona: Lynx Edicions. Hoyo, J. del, Christie, D. A. and Birdlife, I. (2007) Handbook of the birds of the world. Volume 12: Picathartes to Tits and Chickadees. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Christie, D. A. (2003) Handbook of the birds of the world. Volume 8: Broadbills to Tapaculos. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Christie, D. A. (2005) Handbook of the birds of the world. Volume 10: Cuckoo-shrikes to thrushes. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Christie, D. A. (2006) Handbook of the birds of the world. Volume 11: Old world flycatchers to old world warblers. Barcelona: Lynx Edicions. 122

Hoyo, J. del, Elliott, A. and Christie, D. A. (2008) Handbook of the birds of the world. Volume 13: Penduline-tits to shrikes. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Sargatal, J. (1992) Handbook of the birds of the world. Volume 1: Ostrich to ducks. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Sargatal, J. (1996) Handbook of the birds of the world. Volume 3: Hoatzin to auks. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Sargatal, J. (1999) Handbook of the birds of the world. Volume 5: Barnowls to hummingbirds. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Sargatal, J. (2001) Handbook of the birds of the world. Volu,e 6: Mousebirds to hornbills. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Sargatal, J. (2002) Handbook of the birds of the world. Volume 7: Jacamars to woodpeckers. Barcelona: Lynx Edicions. Hoyo, J. del, Elliott, A. and Sargatal, J. (2004) Handbook of the birds of the world. Volu,e 9: Cotingas to pipits and wagtails. Barcelona: Lynx Edicions. Hu, D., Hou L., Zhang L., and Xu X. (2009) ‘A pre-Archaeopteryx troodontid theropod from China with long feathers on the metatarsus’, Nature. 461(7264), 640. Hu, H., O’Connor, J. K. and Zhou, Z. (2015) ‘A new species of Pengornithidae (Aves: Enantiornithes) from the Lower Cretaceous of China suggests a specialized scansorial habitat previously unknown in early birds’, PLoS One. 10(6), e0126791. Ionising Radiation (Medical Exposure) Regulations 2017 (2017). UK: www.legislation.gov.uk. Irwin, R. E. (1996) ‘The phylogenetic content of avian courtship display and song evolution’, Phylogenies and the comparative method in animal behavior. Oxford Univ. Press, Oxford, UK, pp. 234–252. Jarvis, E. D., Mirarab S., Aberer A. J., Li B., Houde P., Li C., Ho s. Y. W., Faircloth B. C., Nabholz B., and Howard J. T. (2014) ‘Whole-genome analyses resolve early branches in the tree of life of modern birds’, Science. 346(6215), 1320–1331. Jombart, T., Balloux, F. and Dray, S. (2010) ‘Adephylo: new tools for investigating the phylogenetic signal in biological traits’, Bioinformatics. 26(15), 1907–1909. Juanes, F. and Hartwick, E. B. (1990) ‘Prey size selection in Dungeness crabs: the effect of claw damage’, Ecology. 71(2), 744–758. Kaestner, A. P., Kis Z., Radebe M. J., Mannes D., Hovind J., Grünzweig C., Kardjilov N., and Lehmann E. H. (2017) ‘Samples to determine the resolution of neutron radiography and tomography’, Physics Procedia. 88, 258–265. Kardong, K. V (2012) ‘Vertebrates: comparative anatomy, function, evolution’, Washingt State Univ, 33, 81–87. Kozub, D., Khmelik V., Shapoval Y., Chentsov V., Yatsenko S., Litovchenko B., and Starykh V. (2008) ‘Helicon Focus Software’. Kuhn, M. (2008) ‘Building predictive models in R using the caret package’, Journal of statistical software, 28(5), 1–26. Lauder, G. V and Thomason, J. J. (1995) ‘On the inference of function from structure’, in Thomason, J. (ed.) Functional morphology in vertebrate paleontology, Cambridge University Press, pp. 1–18. Lautenschlager, S. (2014) ‘Morphological and functional diversity in therizinosaur claws and the implications for theropod claw evolution’, Proceedings of the Royal Society B: Biological Sciences. 281(1785), 20140497. Lee, M. S. Y. and Worthy, T. H. (2011) ‘Likelihood reinstates Archaeopteryx as a primitive bird’, Biology letters. 8(2), 299–303.

123

Lefèvre, U., Hu D., Escuillié F., Dyke G., and Godefroit P. (2014) ‘A new long-tailed basal bird from the Lower Cretaceous of north-eastern China’, Biological Journal of the Linnean Society. 113(3), 790–804. Li, Q., Gao K-Q., Meng Q., Clarke J. A., Shawkey M. D., D’Alba L., Pei R., Ellison M., Norell M. A., and Vinther J. (2012) ‘Reconstruction of Microraptor and the evolution of iridescent plumage’, Science. 335(6073), 1215–1219. Liu, Z., Zhang, Z. and Ritchie, R. O. (2018) ‘On the materials science of nature’s arms race’, Advanced Materials. 30(32), 1705220. Lix L. M., Keselman J. C., Keselman H. J. (1996) ‘Consequences of assumption violations revisited: A quantitative review of alternatives to the one-way analysis of variance F test’, Review of educational research. 66(4), 579-619. Longrich, N. (2006) ‘Structure and function of hindlimb feathers in Archaeopteryx lithographica’, Paleobiology. 32(3), 417–431. Lü, J. and Brusatte, S. L. (2015) ‘A large, short-armed, winged dromaeosaurid (Dinosauria: Theropoda) from the Early Cretaceous of China and its implications for feather evolution’, Scientific Reports. 5, 11775. MacLeod, N. and Rose, K. D. (1993) ‘Inferring locomotor behavior in Paleogene mammals via eigenshape analysis’, American Journal of Science, 293(A), 300. Makdissi, J., Pawar R. P., Johnson B., and Chong B. S. (2016) ‘The effects of device position on the operator’s radiation dose when using a handheld portable X-ray device’, Dentomaxillofacial Radiology. 45(3), 20150245. Mann, S. and Picard, R. (1994) ‘Being ‘undigital’ with digital cameras’, MIT Media Lab Perceptual, 1, 2. Manning, P. L., Margretts L., Johnson M. R., Withers P. J., Sellers W. I., Falkingham P. L., Mummery P. M., Barrett P. M., and Raymont D. R. (2005) ‘Dinosaur killer claws or climbing crampons?’, Biology Letters. 2(1), 110–112. Manning, P. L., Payne D., Pennicott J., Barrett P. M., and Ennos R. A. (2009) ‘Biomechanics of dromaeosaurid dinosaur claws: application of X‐ray microtomography, nanoindentation, and finite element analysis’, The Anatomical Record: Advances in Integrative Anatomy and Evolutionary Biology: Advances in Integrative Anatomy and Evolutionary Biology. 292(9), 1397–1405. Marsh, O. C. (1880) Odontornithes: a monograph on the extinct toothed birds of North America. Museum. Marshall, C. R. (2006) ‘Explaining the Cambrian “explosion” of animals’, Annu. Rev. Earth Planet. Sci. Annual Reviews, 34, pp. 355–384. Mattheck, C. and Reuss, S. (1991) ‘The claw of the tiger: an assessment of its mechanical shape optimization’, Journal of Theoretical Biology. 150(3), 323–328. Mayr, G. and Peters, D. S. (2007) ‘The foot of Archaeopteryx: Response to Feduccia et al. (2007)’, The Auk. 124(4), 1450–1452. Münkemüller, T., Lavergne S., Bzeznik B., Dray S., Jombart T., Schiffers K., and Thuiller W. (2012) ‘How to measure and test phylogenetic signal’, Methods in Ecology and Evolution. 3(4), 743–756. Naish, D. (2000a) ‘130 years of tree-climbing dinosaurs: Archaeopteryx, “arbrosaurs”, and the origin of avian flight’, Quarterly Journal of the Geological Society, London, 4. Naish, D. (2000b) ‘Theropod dinosaurs in the trees: a historical review of arboreal habits amongst nonavian theropods’, Archaeopteryx, 18, 35–41. Norberg, U. M. (1985) ‘Evolution of vertebrate flight: an aerodynamic model for the transition from gliding to active flight’, The American Naturalist. 126(3), 303–327. 124

Norberg, U. M. L. (2007) ‘Evolution of flight in animals’, Flow phenomena in nature, 1, 36– 48. Norell, M. A. and Xu, X. (2005) ‘Feathered dinosaurs’, Annu. Rev. Earth Planet. Sci. Annual Reviews, 33, 277–299. O’Connor, J., Wang X., Sullivan C., Zheng X., Tubaro P., Zhang X., and Zhou Z. (2013) ‘Unique caudal plumage of Jeholornis and complex tail evolution in early birds’, Proceedings of the National Academy of Sciences. 110(43), 17404–17408. O’Connor, J. K. (2019) ‘The trophic habits of early birds’, Palaeogeography, palaeoclimatology, palaeoecology. 513, 178–195. O’Connor, J., Zhou, Z. and Xu, X. (2011) ‘Additional specimen of Microraptor provides unique evidence of dinosaurs preying on birds’, Proceedings of the National Academy of Sciences. 108(49), 19662–19665. Olshevsky, G. (1994) ‘The birds first? A theory to fit the facts’, Omni. General Media International, Inc., 16(9), 34–45. Organ, C. L., Shedlock A. M., Meade A., Pagel M., and Edwards S. V. (2007) ‘Origin of avian genome size and structure in non-avian dinosaurs’, Nature. 446(7132), 180. Osmólska, H. (1987) ‘Borogovia gracilicrus gen. et. sp. n., a new troodontid dinosaur from the Late Cretaceous of Mongolia’, Acta Palaeontologica Polonica. -, 32(1–2). Ostrom, J. H. (1974) ‘Archaeopteryx and the origin of flight’, The Quarterly Review of Biology. 49(1), 27–47. Ostrom, J. H. (1976) ‘Archaeopteryx and the origin of birds’, Biological Journal of the linnean Society. 8(2), 91–182. Ostrom, J. H. (1985) ‘The meaning of Archaeopteryx’, in The beginnings of birds: proceedings of the international Archaeopteryx conference. Eichstätt, Germany, 161– 175. Ostrom, J. H. (1986) ‘The cursorial origin of avian flight’, Mem. Calif. Acad. Sci, 8, 73–81. Padian, K. (1982) ‘Running, leaping, lifting off’, The Sciences. 22(5), 10–15. Padian, K. and Chiappe, L. M. (1998) ‘The origin and early evolution of birds’, Biological reviews. 73(1), 1–42. Parkes, K. C. (1966) ‘Speculations on the origin of feathers’, Living Bird, 5(s 77). Patiño, S. J. and Fariña, R. A. (2017) ‘Ungual phalanges analysis in Pleistocene ground sloths (Xenarthra, Folivora)’, Historical Biology. 29(8), 1065–1075. Paul, G. S. (1988) Predatory dinosaurs of the world: a complete illustrated guide. Simon and Schuster New York. Pavoine, S., Ollier S., Pontier D., and Chessel D. (2008) ‘Testing for phylogenetic signal in phenotypic traits: new matrices of phylogenetic proximities’, Theoretical population biology. 73(1), 79–91. Peters, D. S. and Görgner, E. (1992) ‘A comparative study on the claws of Archaeopteryx’, Papers in avian paleontology honoring Pierce Brodkorb. Edited by KC Campbell Jr. Natural History Museum of Los Angeles County Science Series, 36, 29–37. Petie, R. and Muller, M. (2007) ‘Curvature facilitates prey fixation in predatory insect claws’, Journal of theoretical biology. 244(4), 565–575. Pike, A. V. L. and Maitland, D. P. (2004) ‘Scaling of bird claws’, Journal of Zoology. 262(1), 73–81. Pittayapat, P., Oliveira-Santos C., Thevissen P., Michielsen K., Bergans N., Willems G., Debruyckere D., and Jacobs R. (2010) ‘Image quality assessment and medical physics evaluation of different portable dental X-ray units’, Forensic science international. 201(1–3), 112–117.

125

Pongnapang, N. (2005) ‘Practical guidelines for radiographers to improve computed radiography image quality’, Biomedical imaging and intervention journal. 1(2). Popović, Z. B. and Thomas, J. D. (2017) ‘Assessing observer variability: a user’s guide’, Cardiovascular diagnosis and therapy. 7(3), 317. Prasad, S. and Bruce, L. M. (2008) ‘Limitations of principal components analysis for hyperspectral target recognition’, IEEE Geoscience and Remote Sensing Letters. 5(4), 625–629. Prum, R. O., Berv J. S., Dornburg A., Field D. J., Townsend J. P., Lemmon E. M., and Lemmon A. R. (2015) ‘A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing’, Nature. 526(7574), 569. Pu, H., Chang H., Lü J., Wu Y., Xu L., Zhang J., and Jia S. (2013) ‘A new juvenile specimen of Sapeornis (Pygostylia: Aves) from the Lower Cretaceous of Northeast China and allometric scaling of this basal bird’, Paleontological Research. 17(1), 27–39. Rauhut, O. W. M., Foth, C. and Tischlinger, H. (2018) ‘The oldest Archaeopteryx (Theropoda: Avialiae): a new specimen from the Kimmeridgian/Tithonian boundary of Schamhaupten, Bavaria’, PeerJ. 6, e4191. Ribas, S. C., Velloso A. L. R., Teixeira-Filho P., Rocha-Barbosa O., Evangelista H., and Dos Santos E. A. (2004) ‘Structure of claws and toes of two tropidurid lizard species of Restinga from Southeastern Brazil: adaptations to the vertical use of the habitat’, Revista Chilena de Historia Natural, 77(4), 599–606. Rothschild, B. M., Bryant B., Hubbard C., Tuxhorn K., Kilgore G. P., Martin L., and Naples V. (2013) ‘The power of the claw’, PloS one. 8(9), e73811. Rubin, C. T. and Lanyon, L. E. (1982) ‘Limb mechanics as a function of speed and gait: a study of functional strains in the radius and tibia of horse and dog’, Journal of experimental biology. 101(1), 187–211. Senter, P., Kirkland J. I., DeBlieux D. D., Madsen S., and Toth N. (2012) ‘New dromaeosaurids (Dinosauria: Theropoda) from the Lower Cretaceous of Utah, and the evolution of the dromaeosaurid tail’, PLoS One. 7(5), e36790. Shen, C., Zhao B., Gao C., Lü J., and Kundrát M. (2017) ‘A new troodontid dinosaur (Liaoningvenator curriei gen. et sp. nov.) from the Early Cretaceous Yixian Formation in western Liaoning province’, Acta Geoscientica Sinica, 38(3), 359–371. Siu, D., Cooke T. D., Broekhoven L. D., Lam M., Fisher B., Saunders G., and Challis T. W. (1991) ‘A standardized technique for lower limb radiography. Practice, applications, and error analysis.’, Investigative radiology, 26(1), 71–77. Spielmann, J. A., Heckert, A. B. and Lucas, S. G. (2005) ‘The Late Triassic archosauromorph Trilophosaurus as an arboreal climber’, Rivista Italiana di Paleontologia e Stratigrafia (Research In Paleontology and Stratigraphy), 111(3). Spielmann, J. A., Renesto, S. and Lucas, S. G. (2006) ‘THE UTILITY OF CLAW CURVATURE IN ASSESSING THE ARBOREALITY OF FOSSIL REPTILES’, New Mexico Museum of Natural History and Science Bulletin, 37, 365–368. Stevens, K. A. (2006) ‘Binocular vision in theropod dinosaurs’, Journal of Vertebrate Paleontology. 26(2), 321–330. Sullivan, C., Wang Y., Hone D. W. E., Wang Y., Xu X., and Zhang F. (2014) ‘The vertebrates of the Jurassic Daohugou Biota of northeastern China’, Journal of Vertebrate Paleontology. 34(2), 243–280. Sustaita, D. and Hertel, F. (2010) ‘In vivo bite and grip forces, morphology and prey-killing behavior of North American accipiters (Accipitridae) and falcons (Falconidae)’, Journal of Experimental Biology. 213(15), 2617–2628.

126

Takahashi, A. S. and Sakuma, S. (2012) Magnification radiography. Springer Science & Business Media. Team, R. C. (2013) ‘R: A language and environment for statistical computing’. Vienna, Austria. Thompson, D. W. (1942) On Growth and Form, Second Edition. Cambridge University Press. Tinius, A. and Russell, A. (2017) ‘Points on the curve: An analysis of methods for assessing the shape of vertebrate claws’, Journal of morphology. 278(2), 150–169. Tsang, L. R., Wilson L. A. B., Ledogar J., Wroe S., Attard M., and Sansalone G. (2019) ‘Raptor talon shape and biomechanical performance are controlled by relative prey size but not by allometry’, Scientific reports. 9(1), 7076. Tsang, L. R., Wilson, L. A. B. and McDonald, P. G. (2019) ‘Comparing the toepads of Australian diurnal and nocturnal raptors with nonpredatory taxa: Insights into functional morphology’, Journal of morphology. 280(11), 1682-1692. Tsuihiji, T., Barsbold R., Watabe M., Tsogtbaatar K., Suzuki S., and Hattori S. (2016) ‘New material of a troodontid theropod (Dinosauria: Saurischia) from the Lower Cretaceous of Mongolia’, Historical Biology. 28(1–2), 128–138. Tulli, M. J., Cruz F. B., Herrel A., Vanhooydonck B., and Abdala V. (2009) ‘The interplay between claw morphology and microhabitat use in neotropical iguanian lizards’, Zoology. 112(5), 379–392. Tulli, M. J., Abdala, V. and Cruz, F. B. (2011) ‘Relationships among morphology, clinging performance and habitat use in Liolaemini lizards’, Journal of evolutionary biology. 24(4), 843–855. Turner, A. H., Pol D., Clarke J. A., Erickson G. M., and Norell M. A. (2007) ‘A basal dromaeosaurid and size evolution preceding avian flight’, Science. 317(5843), 1378– 1381. Turner, A. H., Makovicky, P. J. and Norell, M. A. (2007) ‘Feather quill knobs in the dinosaur Velociraptor’, Science. 317(5845), 1721. Turner, A. H., Makovicky, P. J. and Norell, M. A. (2012) ‘A review of dromaeosaurid systematics and paravian phylogeny’, Bulletin of the American museum of natural history. 2012(371), 1–206. Van Valkenburgh, B. (1987) ‘Skeletal indicators of locomotor behavior in living and extinct carnivores’, Journal of Vertebrate Paleontology. 7(2), 162–182. Wainwright, P. C., Alfaro M. E., Bolnick D. I., and Hulsey C. D. (2005) ‘Many-to-one mapping of form to function: a general principle in organismal design?’, Integrative and comparative biology. 45(2), 256–262. Wang, M., Zhou Z., O’Connor J. K., and Zelenkov N. V. (2014) ‘A new diverse enantiornithine family (Bohaiornithidae fam. nov.) from the Lower Cretaceous of China with information from two new species’, Vertebr. Palasiat, 52, 31–76. Wang, M., O’Connor J. K., Xu X., and Zhou Z. (2019) ‘A new Jurassic scansoriopterygid and the loss of membranous wings in theropod dinosaurs’, Nature. 569(7755), 256. Wang, M., O’Connor, J. K. and Zhou, Z. (2014) ‘A new robust enantiornithine bird from the Lower Cretaceous of China with scansorial adaptations’, Journal of Vertebrate Paleontology. 34(3), 657–671. Wang, X., O’Connor J. K., Zheng X., Wang M., Hu H., and Zhou Z. (2014) ‘Insights into the evolution of rachis dominated tail feathers from a new basal enantiornithine (Aves: Ornithothoraces)’, Biological Journal of the Linnean Society. 113(3), 805–819. Warton, D. I., Duursma R. A., Falster D. S., and Taskinen S. (2012) ‘smatr 3–an R package for estimation and inference about allometric lines’, Methods in Ecology and Evolution. 3(2), 257–259. 127

Wickham, H., Francois R., Henry L., and Müller K. (2015) ‘dplyr: A grammar of data manipulation’, R package version 0.4, 3. Wickham, H. (2016) ggplot2: elegant graphics for data analysis. Springer. Witmer, L. M. and Ridgely, R. C. (2009) ‘New insights into the brain, braincase, and ear region of tyrannosaurs (Dinosauria, Theropoda), with implications for sensory organization and behavior’, The Anatomical Record: Advances in Integrative Anatomy and Evolutionary Biology. 292(9), 1266–1296. Witmer, L. M. and Thomason, J. J. (1995) ‘The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils’, Functional morphology in vertebrate paleontology, 1, 19–33. Wolff, J. (1986) ‘The law of bone remodeling’, Berlin: Heidelberg & New York: Springer, pp. 63–65. Xing, L., Persons IV W. S., Bell P. R., Xu X., Zhang J., Miyashita T., Wang F., and Currie P. J. (2013) ‘Piscivory in the feathered dinosaur Microraptor’, Evolution. 67(8), 2441– 2445. Xu, X., Zhou Z., Wang X., Kuang X., Zhang F., and Du X. (2003) ‘Four-winged dinosaurs from China’, Nature. 421(6921), 335. Xu, X., Zhao Q., Norell M., Sullivan C., Hone D., Erickson G., Wang X., Han F., and Guo Y. (2009) ‘A new feathered maniraptoran dinosaur fossil that fills a morphological gap in avian origin’, Chinese Science Bulletin. 54(3), 430–435. Xu, X., You H., Du K., and Han F. (2011) ‘An Archaeopteryx-like theropod from China and the origin of Avialae’, Nature. 475(7357), 465. Xu, X., Zhou Z., Dudley R., Mackem S., Chuong C-M., Erickson G. M., and Varricchio D. J. (2014) ‘An integrative approach to understanding bird origins’, Science. 346(6215), 1253293. Xu, X., Zheng X., Sullivan C., Wang X., Xing L., Wang Y., Zhang X., O’Connor J. K., Zhang F., and Pan Y. (2015) ‘A bizarre Jurassic maniraptoran theropod with preserved evidence of membranous wings’, Nature. 521(7550), 70. Xu, X. and Zhang, F. (2005) ‘A new maniraptoran dinosaur from China with long feathers on the metatarsus’, Naturwissenschaften. 92(4), 173–177. Xu, X., Zhou, Z. and Wang, X. (2000) ‘The smallest known non-avian theropod dinosaur’, Nature. 408(6813), 705. Yalden, D. W. (1985) ‘Forelimb function in Archaeopteryx’, The beginnings of birds. Eichstätt Freunde des Jura-Museums Eichstätt, pp. 91–97. Yalden, D. W. (1997) ‘Climbing Archaeopteryx’, Archaeopteryx, 15, 107–108. Yongdong, W., Ken-ichi S., Wu Z., and Shaolin Z. (2006) ‘Biodiversity and palaeoclimate of the Middle Jurassic floras from the Tiaojishan Formation in western Liaoning, China’, Progress in Natural Science. 16(sup1), 222–230. Yuri, T., Kimball R., Harshman J., Bowie R., Braun M., Chojnowski J., Han K-L., Hackett S., Huddleston C., and Moore W. (2013) ‘Parsimony and model-based analyses of indels in avian nuclear genes reveal congruent and incongruent phylogenetic signals’, Biology. 2(1), 419–444. Zani, P. A. (2000) ‘The comparative evolution of lizard claw and toe morphology and clinging performance’, Journal of evolutionary biology, 13(2), 316–325. Zanno, L. E., Varricchio D. J., O’Connor P. M., Titus A. L., and Knell M. J. (2011) ‘A new troodontid theropod, Talos sampsoni gen. et sp. nov., from the Upper Cretaceous Western Interior Basin of North America’, PloS one. 6(9), e24487. Zelenitsky, D. K. (2006) ‘Reproductive traits of non-avian theropods’, Journal of the Paleontological Society of Korea. 22(1), 209. 128

Zhang, F., Zhou Z., Xu X., Wang X., and Sullivan C. (2008) ‘A bizarre Jurassic maniraptoran from China with elongate ribbon-like feathers’, Nature. 455(7216), 1105. Zhang, Y., O’Connor J., Di L., Qingjin M., Sigurdsen T., and Chiappe L. M. (2014) ‘New information on the anatomy of the Chinese Early Cretaceous Bohaiornithidae (Aves: Enantiornithes) from a subadult specimen of Zhouornis hani’, PeerJ. 2, e407. Zhang, Z., Gao C., Meng Q., Liu J., Hou L., and Zheng G. (2009) ‘Diversification in an Early Cretaceous avian genus: evidence from a new species of Confuciusornis from China’, Journal of Ornithology. 150(4), 783–790. Zheng, X., Zhou Z., Wang X., Zhang F., Zhang X., Wang Y., Wei G., Wang S., and Xu X. (2013) ‘Hind wings in basal birds and the evolution of leg feathers’, Science. 339(6125), 1309–1312. Zhou, Z.-H. and Farlow, J. O. (2001) ‘Flight capability and habits of Confuciusornis’, New perspectives on the origin and early evolution of birds. pp. 237–254. Zhou, Z., Barrett, P. M. and Hilton, J. (2003) ‘An exceptionally preserved Lower Cretaceous ecosystem’, Nature. 421(6925), 807. Zhou, Z. and Zhang, F. (2002) ‘Largest bird from the Early Cretaceous and its implications for the earliest avian ecological diversification’, Naturwissenschaften. 89(1), 34–38. Zhou, Z. and Zhang, F. (2003) ‘Jeholornis compared to Archaeopteryx, with a new understanding of the earliest avian evolution’, Naturwissenschaften. 90(5), 220–225. Zinoviev, A. V (2009) ‘An attempt to reconstruct the lifestyle of confuciusornithids (Aves, Confuciusornithiformes)’, Paleontological Journal. 43(4), 444–452.

129

Appendix

A. Radiographs

Figure A.1: Accipiters. From left to right: (Top row) Accipiter gentilis specimen 1969-228-7, Accipiter nisus unregistered specimen, Buteo buteo (right claw) unregistered specimen, Buteo buteo (left claw) unregistered specimen. (Middle row) Circaetus gallicus specimen 10444, Milvus migrans specimen 7960, Pernis apivoris specimen 10497. Bottom row from left to right: (Bottom row) Cathartes aura specimen 8360, Pandion haliaetus specimen D.296b

Figure A.2: Falcons. From left to right: (Top row) Falco columbarius specimen 10561, Falco rusticolus specimen 242b, Falco tinnunculus specimen BB.6156, Falco peregrinus specimen 10561. (Bottom row) Phalcoboenas megalopterus specimen 228-175.

130

Figure A.3: Strigiformes. From left to right: (Top row) Bubo virginianus specimen 1986.2.188, Strix aluco specimen 10699, Strix aluco (left claw) unregistered specimen, Strix aluco (right claw) unregistered specimen. (Middle row) Strix uralensis specimen 10708, Tyto alba (left claw) unregistered specimen, Tyto alba (right claw) unregistered specimen, Tyto novaehollandiae specimen 285. (Bottom row) Tyto tenebricosa specimen B578

Figure A.4: Palaeognathae. From left to right: (Top row) Apteryx owenii specimen 8380, Crypturellus obsoletus specimen 2960, Rhynchotus rufescens specimen 4646. (Bottom row) Eudromia elegans (left claw) specimen 4621, Eudromia elegans (right claw) specimen 4621, Tinamus major specimen 3650a, Tinamus major specimen 1923.20.16.

Figure A.5: Otidiformes. From left to right: Eupodotis senegalensis unregistered specimen, Tetrax tetrax specimen 8625, Tetrax tetrax specimen 1956.

131

Figure A.6: Water birds. From left to right: Numenius gravata specimen 1980-281, Larus canus specimen 1384, Ixobrychus falvicollis specimen B.602, Calonectris diomedea specimen Bb.84.72.

Figure A.7: Cuculiformes. From left to right: (Top row) Centropus phasianus specimen 1999.20.71, Centropus menbeki specimen 1174, Cuculus canorus specimen BD10281 exposure .25, Cuculus canorus specimen BD10281 exposure .28. (Bottom row) Piaya cayana specimen BD10960 exposure .18, Piaya cayana specimen BD10960 exposure .19.

Figure A.8: Galliformes. From left to right: (Top row) Alectoris chukar specimen 121a, Lagopus lagopus specimen 1984.2.197, Phasianus colchicus (left claw) unregistered specimen, Phasianus colchicus (right claw) unregistered specimen. 132

(Middle row) Syrmaticus soemmerringii specimen 16514, Tetrao tetrix specimen 1982, Tragopan blythii specimen 1984.2.254, Megapodius freycinet specimen 3423a. (Bottom row) Ortalis superciliaris specimen 1644, Ortalis vetula specimen 488a.

Figure A.9: Gruiformes. From left to right: Aramus guarauna specimen Bb.71.97, Psophia leucoptera specimen 1984.2.294, Psophia viridis specimen 253.

Figure A.10: Parrots. From left to right: (Top row) Nestor notabilis specimen 2522, Strigops habroptilus specimen 18329, Cacatua sulphurea specimen T1180, Eolophus roseicapillus specimen B.68. (Second row) Ara macao specimen 25.8.09, Loriculus vernalis specimen 3510 exposure .21, Loriculus vernalis specimen 3510 exposure .12, Loriculus vernalis specimen 3510 exposure .14, (Third row) Loriculus vernalis specimen 3510 exposure .19, Loriculus vernalis specimen 3510 exposure .2, Loriculus vernalis specimen 3510 exposure .25. (Bottom row) Alisterus scapularis specimen 1966.140.87, Alisterus scapularis specimen B.48. Trichoglossus haematodus unregistered specimen

133

Figure A.11: Hornbills, rollers, and kingfishers. From left to right: (Top row) Tockus nasutus specimen 29-9-1899, Dacelo novaeguineae specimen 140.1.46, Megaceryle maxima specimen 1920.146.7, Eurystomus glaucurus specimen 10854. (Bottom row) Leptosomus discolour specimen 16111.

Figure A.12: Musophagiformes, Caprimulgiformes, Pterocliformes, Opisthocomiformes. From left to right: Crinifer zonurus specimen 6265, Steatornis caripensis specimen 1223, Pterocles orientalis specimen 578e, Opisthocomus hoazin specimen 12729.

Figure A.13: Columbiformes. From left to right: (Top row) Columba oenas specimen 8083, Columba palumbus (right claw) unregistered specimen, Columba palumbus (left claw) unregistered specimen, Ducula poliocephala specimen 1985.7.136. (Bottom row) Hemiphaga novaeseelandiae specimen 3544d, Lopholaimus antarcticus specimen 4213.

134

Figure A.14: Corvids. From left to right: (Top row) Corvus corax specimen Bb.7043, Corvus corone (right claw) unregistered specimen, Corvus corone (left claw) unregistered specimen, Garrulus glandarius specimen 1970.76.850. (Bottom row) Nucifraga caryocatactes specimen 12000.

135

Figure A.15: Woodcreepers. From left to right: (Top row) Dendrocolaptes certhia specimen 8.1910.14 exposure .14, Dendrocolaptes certhia specimen 8.1910.14 exposure .15, Dendrocolaptes certhia specimen 8.1910.14 exposure .16. (Second row) Dendrocolaptes certhia specimen 8.1910.14 exposure .17, Dendrocolaptes certhia specimen 8.1910.14 exposure .18, Dendrocolaptes certhia specimen 8.1910.14 exposure .19, Dendrocolaptes certhia specimen 8.1910.14 exposure .2. (Third row) Dendrocolaptes certhia specimen 8.1910.14 exposure .21, Dendrocolaptes certhia specimen 8.1910.14 exposure .22, Dendrocolaptes certhia specimen 8.1910.14 exposure .23, Dendrocolaptes picumnus specimen 6196. (Fourth row) Nasica longirostris specimen 1332S exposure .18, Nasica longirostris specimen 1332S exposure .19, Nasica longirostris specimen 1332S exposure .2, Nasica longirostris specimen 1332S exposure .21. (Fifth row) Nasica longirostris specimen 1332S exposure .22, Nasica longirostris specimen 1332S exposure .23, Nasica longirostris specimen 1332S exposure .25, Xiphocolaptes albicollis specimen D.14.4b,a. (Sixth row) Xiphocolaptes major specimen D.1427 exposure .14, Xiphocolaptes major specimen D.1427 exposure .15, Xiphocolaptes major specimen D.1427 exposure .16, Xiphocolaptes major specimen D.1427 exposure .17. (Seventh row) Xiphocolaptes major specimen D.1427 exposure .18, Xiphocolaptes major specimen D.1427 exposure .19, Xiphocolaptes major specimen D.1427 exposure .2, Xiphocolaptes major specimen D.1427 exposure .21. (Eighth row) Xiphocolaptes major specimen D.1427 exposure .22, Xiphocolaptes major specimen D.1427 exposure .23, Xiphocolaptes major specimen D.1427 exposure .24, Xiphocolaptes major specimen D.1427 exposure .25.

136

(Ninth row) Xiphocolaptes promeropirhynchus specimen D.1430 exposure .16, Xiphocolaptes promeropirhynchus specimen D.1430 exposure .17, Xiphocolaptes promeropirhynchus specimen D.1430 exposure .18, Xiphocolaptes promeropirhynchus specimen D.1430 exposure .19. (Bottom row) Xiphocolaptes promeropirhynchus specimen D.1430 exposure .2, Xiphocolaptes promeropirhynchus specimen D.1430 exposure .22, Xiphocolaptes promeropirhynchus specimen D.1430 exposure .23, Xiphocolaptes promeropirhynchus specimen D.1430 exposure .24.

Figure A.16: Toucans. From left to right: (Top row) Ramphastos toco specimen Bb7896, Ramphastos tucanus specimen 13.10.04.22, Selenidera maculirostris specimen 77a exposure .15, Selenidera maculirostris specimen 77a exposure .16. (Middle row) Selenidera maculirostris specimen 77a exposure .17, Selenidera maculirostris specimen 77a exposure .18, Selenidera maculirostris specimen 77a exposure .19, Selenidera maculirostris specimen 77a exposure .21. (Bottom row) Selenidera maculirostris specimen 77a exposure .22, Selenidera maculirostris specimen 77a exposure .23, Selenidera maculirostris specimen 77a exposure .24, Selenidera maculirostris specimen 77a exposure .25.

137

Figure A.17: Woodpeckers and barbets. From left to right: (Top row) Campephilus magellanicus specimen 3777, Campephilus principalis unregistered specimen, Colaptes auratus specimen BB.7122, Colaptes campestris specimen B.4384. (Second row) Dryocopus javensis specimen 3855, Dryocopus martius specimen B.5986, Dryocopus martius specimen 3864, Dryocopus pileatus specimen 7116. (Third row) Mulleripicus pulverulentus specimen 4112, Mulleripicus pulverulentus specimen 1978, Picus squamatus specimen B.11308, Picus viridis specimen 1324a. (Bottom row) Megalaima chrysopogon specimen BD10010 exposure .27, Megalaima chrysopogon specimen BD10010 exposure .28, Megalaima chrysopogon specimen BD10010 exposure .29, Psilopogon lineatus specimen BD10206.

Figure A.18: Squamates. From left to right: (Top row) Chamaeleo melleri specimen 60.1 exposure .18, Chamaeleo melleri specimen 60.1 exposure .2, Ctenosaura acanthura specimen 117 exposure .16, Ctenosaura acanthura specimen 117 exposure .17. (Middle row) Tiliqua scincoides specimen 1867-1, Uromastyx hardwickii specimen 1987.143.55 exposure .15, Uromastyx hardwickii specimen 1987.143.55 exposure .16, Uromastyx hardwickii specimen 1987.143.55 exposure .17. (Bottom row) Varanus gouldii specimen 2000.1.32 exposure .15, Varanus gouldii specimen 2000.1.32 exposure .18.

138

B. Photographs of fossil claws

Figure B.1: Avialans. From left to right: (Top row) 12th specimen of Archaeopteryx right claw (Rauhut et al. 2018), 12th specimen of Archaeopteryx left claw (Rauhut et al. 2018), Berlin specimen of Archaeopteryx left claw (MFN), Berlin specimen of Archaeopteryx right claw (MFN)

139

(Second row) Balaur bondoc specimen EME PV.313 (Turner et al. 2012), Confuciusornis sanctus specimen IVPP V 13156 (Falk et al. 2016), Confuciusornis sanctus specimen MB.Av.1158 (MFN) (Third row) Confuciusornis sanctus specimen IVPP V 14412 A (right claw) (Wang et al. 2018), Confuciusornis sanctus specimen IVPP V 14412 A (left claw) (Wang et al. 2018), Eopengornis specimen STM24-1 (Wang et al. 2014), Fortunguavis xiaotaizicus specimen IVPP V18631 (Wang et al. 2014) (Fourth row) Parapengornis specimen IVPP V18632 (Hu et al. 2015), Sapeornis specimen STM16-18 (Zheng et al. 2013), Sapeornis specimen HGM-41HIII0405 (left claw) (Pu et al. 2013), Sapeornis specimen HGM-41HIII0405 (right claw) (Pu et al. 2013) (Bottom row) Sapeornis specimen DNHM-D3078 (Gao et al. 2012), Zhouornis hani specimen BMNHC Ph 756 (right claw) (Zhang et al. 2014), Zhouornis hani specimen BMNHC Ph 756 (left claw) (Zhang et al. 2014)

Figure B.2: Dromaeosaurids. From left to right: (Top row) Changyuraptor yangi specimen HG B016 (Han et al. 2014), Microraptor zhaoianus specimen CAGS 20-8-001 (right claw) (Hwang et al., 2002), Microraptor zhaoianus specimen CAGS 20-8-001 (left claw) (Hwang et al., 2002) (Bottom row) Halszkaraptor specimen MPC D-102/109 (Cau et al., 2017), Zhenyuanlong specimen JPM-0008 (Lü & Brusatte 2015)

140

Figure B.3: Troodontids. From left to right: Borogovia gracilicrus specimen ZPAL MgD- I/174 (Osmólska, 1987), specimen MPC-D 100/140 (Tsuihiji et al., 2016), Talos sampsoni specimen UMNH VP 19479 (Zanno et al. 2011)

Figure B.4: Anchiornithids. From left to right: (Top row) Anchiornis huxleyi specimen PKUP V1068 (Pei et al. 2017), Anchiornis huxleyi specimen LPM-B00169 (Hu et al. 2009), Anchiornis huxleyi specimen BMNHC PH823 (Pei et al. 2017) (Bottom row) Eosinopteryx brevipenna specimen YFGP-T5197 (Godefroit et al. 2013), Pedopenna daohugouensis specimen IVPP V12721 (Sullivan et al., 2014), Xiaotingia zhengi specimen STM 27-2 (Xu et al. 2011)

Figure B.5: Ornithomimidae, Tyrannosauridae. From left to right: Beishanlong grandis specimen FRDC-GS JB(07)01-01 (Makovicky et al. 2009), Tyrannosaurus rex specimen FMNH PR2081 (“Sue”) (Brochu, 2003)

141