Primary feather lengths may not be important for inferring the flight styles of Mesozoic birds NICHOLAS R. CHAN, GARETH J. DYKE AND MICHAEL J. BENTON Chan, N.R., Dyke, G.J. & Benton, M.J. 2013: Primary feather lengths may not be impor- tant for inferring the flight styles of Mesozoic birds. Lethaia,Vol.46,pp.146–152. Although many Mesozoic fossil birds have been found with primary feathers preserved, these structures have rarely been included in morphometric analyses. This is surprising because the flight feathers of modern birds can contribute approximately 50% of the total wing length, and so it would be assumed that their inclusion or exclusion would modify functional interpretations. Here we show, contrary to earlier work, that this may not be the case. Using forelimb measurements and primary feather lengths from Meso- zoic birds, we constructed morphospaces for different clades, which we then compared with morphospaces constructed for extant taxa classified according to flight mode. Con- sistent with older work, our results indicate that among extant birds some functional flight groups can be distinguished on the basis of their body sizes and that variation in the relative proportions of the wing elements is conservative. Mesozoic birds, on the other hand, show variable proportions of wing bones, with primary feather length contri- bution to the wing reduced in the earlier diverging groups. We show that the diverse Mesozoic avian clade Enantiornithes overlaps substantially with extant taxa in both size and limb element proportions, confirming previous morphometric results based on skel- etal elements alone. However, these measurements cannot be used to distinguish flight modes in extant birds, and so cannot be used to infer flight mode in fossil forms. Our analyses suggest that more data from fossil birds, combined with accurate functional determination of the flight styles of living forms is required if we are to be able to predict the flight modes of extinct birds. h Birds, flight, morphospace, Mesozoic, wing. Nicholas R. Chan [[email protected]], Department of Biological Sciences, Macquarie University, 2109 New South Wales, Australia; Gareth J. Dyke [gareth.dyke@so- ton.ac.uk], Ocean and Earth Sciences, University of Southampton, Southampton SO14 3ZH UK; Michael Benton [[email protected]], School of Earth Sciences, University of Bristol, Bristol, BS8 1RJ UK; manuscript received on 28 ⁄ 04 ⁄ 12; manuscript accepted on 13 ⁄ 06 ⁄ 12. Modern birds (Neornithes) are the most diverse also likely to have been important to the aerial abilities extant tetrapod group, comprising over 10 400 species of extinct taxa. However, previous attempts to com- (Newton 2003; Gill 2007) and occupying a wide array pare flight-linked aspects of the forelimb between liv- of ecological niches (del Hoyo et al. 1992–2010; Gill ing and extinct birds from the Mesozoic have focused 2007). The ecological diversity of modern birds means on the proportions of the skeletal elements (see McGo- that they have evolved a huge range of flight tech- wan & Dyke 2007; Dyke & Nudds 2009), omitting evi- niques, ranging from short bursts for rapid take-off to dence from thousands of Cretaceous bird fossils that sustained flapping and soaring (Dial 2003; Gill 2007; preserve the remains, or unequivocal imprints, of Bruderer et al. 2010). The many flight modes and flight feathers (but see Wang et al. 2011a, b). speeds in living birds have been shown to be linked to Several studies comparing Mesozoic and extant variations in the morphology of the skeletal and non- birds have previously been carried out. One of these skeletal (e.g. muscles, feathers) elements that comprise compared the forelimb proportions of Enantiornithes and power the wings (Dial & Biewener 1993; Tobalske to those of extant Neornithes, focusing on the propor- et al. 2003; Dawson 2005; Nudds et al. 2007), as well tions of the skeletal wing composed of the three main as to variations in wing loading (weight divided by elements, humerus, ulna and manus (Dyke & Nudds wing area) and aspect ratio (wing span divided by 2009), while others have attempted to compare wing average wing chord) (Rayner 1988). These latter two bone and feather proportions between fossil birds and factors in particular relate to flight speed and manoeu- their living counterparts (Wang et al. 2011a, b). Dif- vrability in all extant flying vertebrates (Norberg & ferences in the lengths of the longest primary feather Rayner 1987; Alerstam et al. 2007). relative to total arm length (ta = humerus + ulna + It is reasonable to argue that morphological charac- manus) in Mesozoic and extant birds have been used ters that are linked to flight mode in living birds are to suggest that the abundant Chinese Cretaceous bird DOI 10.1111/j.1502-3931.2012.00325.x Ó 2012 The Authors, Lethaia Ó 2012 The Lethaia Foundation LETHAIA 46 (2013) Mesozoic bird flight 147 Confuciusornis (Confuciusornithidae) had a different the flight mode is known, rather than large groupings flight mode from extant birds and may well have been of extant birds, as in a previous study (Dyke & Nudds limited to gliding (Wang et al. 2011a, b). This conclu- 2009). Measurements from four specimens consisting sion is also consistent with shoulder anatomy (Rayner of detached wings (three from Turdus merula and one 2001; Senter 2006) as well as the relative lengths and from Buteo buteo) provided by Bristol City Museum rachis widths of its primary feathers (Nudds & Dyke & Art Gallery were added to the original data set 2010; Wang et al. 2011b). Several of these studies have (Nudds et al. 2011), expanding the extant data set to also suggested that the enantiornithines as a group 82 specimens from 34 species. probably encompassed a range of flight modes, including at least some used by modern birds. We also Analyses know that the relative lengths of the primary feathers are important in determining flight style in extant Prior to carrying out principal components analysis birds (Nudds 2007; Nudds et al. 2011). (PCA), the wing was divided into four elements: In this article, we take a new perspective on this humerus(HUM),forearm(FORE),manus(MAN), issue by combining measurements of primary feathers and longest primary feather (PRI). The lengths in mm and the skeletal arm to investigate further the func- of each were log10 transformed and entered into the tional forelimb morphospaces occupied by Mesozoic formula and extant birds (see also Wang et al. 2011a, b). Spe- cifically we test whether, or not, hypotheses that GMw ¼ðlgHUM  lgFORE  lgMAN  lgPRIÞ0:25 invoke similarities in flight mode, or differences in the relative proportions of wing bone lengths can still be supported when primary feathers are included in an GMw is thus an average of the log10 elements, and analysis. log10GMw (lgGMw) was subtracted from the log of each element. This method reduces the amount of variation attributable to size by removing a measure Materials and methods of difference between the actual length of each element and an idealized length (the length of each element if all elements in the arm were equal). As average length Data of each element increases with increasing wing length, Wing measurements are derived from two sources, for so too does GMw and lgGMw, making this idealized extant birds (Nudds et al. 2011) and Mesozoic birds length a measure of overall wing length. The results (Wang et al. 2011b). Distinct from other studies (see were subjected to PCA using PAST (Hammer et al. Wang et al. 2011a), we made the two data sets directly 2001). comparable by including only those Mesozoic birds Each fossil specimen was treated as a separate data for which digit measurements were available: this point. Although we are aware that this artificially reduced our data set to 50 specimens. In addition, inflates sample size, this was necessary as many Meso- measurements for three specimens of Archaeopteryx zoic fossil birds have yet to be fully described and (Munich, London and Maxberg) (Wellnhofer 2009) therefore have not been assigned to any particular spe- were added, and three non-avian theropods with wing cies or genus (Wang et al. 2011b). All specimens feathers were used as an outgroup: Tianyuraptor ostr- awaiting description and included here are identified omi, Microraptor gui and Anchiornis huxleyi (Hu et al. as indeterminate Enantiornithes. Measurements of 2009). specimens of extant taxa were converted to mean val- ues for each species. To test for the robustness of the data entered into Flight modes the PCA against any phylogenetic effects, we then As in previous studies (Nudds et al. 2007; Bruderer employed comparisons by independent contrasts et al. 2010; Wang et al. 2011b), extant birds were (CAIC; Purvis & Rambaut 1995). All taxa in the data- divided into broad, qualitative categories that describe set were placed in the phylogeny of Livezey & Zusi their flight mode. Here, we divided extant taxa into (2007) with all branch lengths considered equal. Sev- the following categories; divers, fast flappers, manoeu- eral of the Mesozoic taxa in our dataset are not vring flappers, slow flappers with rapid take-off, and included in the Livezey & Zusi (2007) study, but we soaring birds, using a standard and widely referenced were able to position them phylogenetically (see source (del Hoyo et al. 1992–2010). This allows direct Fig. 1) according to well resolved published tree comparisons between the forelimb proportions of topologies (e.g. Chiappe & Dyke 2002; O’Connor Mesozoic birds and their extant counterparts where et al. 2009; Zhou & Li 2010). The specimens identified 148 Chan et al. LETHAIA 46 (2013) Table 1. Percentages of variance in the data accounted for by each principal component. lgGMw GMw PC1 88.4 58.7 PC2 6.8 35.4 PC3 4.2 5.8 All four variables load negatively on PC1, indicating that this is a size axis with size decreasing to the right (Fig.