Comparative anatomy of the passerine carpometacarpus helps illuminate the early fossil record of crown Passeriformes

Abstract The hyper‐diverse clade Passeriformes (crown group passerines) comprises over half of extant bird diversity, yet disproportionately few studies have targeted passerine comparative anatomy on a broad phylogenetic scale. This general lack of research attention hinders efforts to interpret the passerine fossil record and obscures patterns of morphological evolution across one of the most diverse clades of extant vertebrates. Numerous potentially important crown passeriform fossils have proven challenging to place phylogenetically, due in part to a paucity of phylogenetically informative characters from across the passerine skeleton. Here, we present a detailed analysis of the morphology of extant passerine carpometacarpi, which are relatively abundant components of the passerine fossil record. We sampled >70% of extant family‐level passerine clades (132 extant species) as well as several fossils from the Oligocene of Europe and scored them for 54 phylogenetically informative carpometacarpus characters optimised on a recently published phylogenomic scaffold. We document a considerable amount of previously undescribed morphological variation among passerine carpometacarpi, and, despite high levels of homoplasy, our results support the presence of representatives of both crown Passeri and crown Tyranni in Europe during the Oligocene.


| INTRODUC TI ON
Passerine birds (Passeriformes) comprise over 6000 extant species, exhibiting a near-global distribution and exceptional phenotypic, ecological and behavioural disparity. As a result of their extraordinary taxonomic diversity, the evolutionary history of extant passerines has frequently been the subject of largescale macroevolutionary studies (e.g., Barker et al., 2004;Day et al., 2020;Derryberry et al., 2011;Moyle et al., 2016;Navalón et al., 2020). However, despite longstanding ornithological interest in passerine evolution, comprehensive characterisations of variation in skeletal morphology across Passeriformes are generally lacking, with most osteological studies on passerines incorporating a scant taxon sample for such a speciose group.
This hinders our understanding of major patterns of anatomical evolution across one of the most conspicuous and diverse extant vertebrate clades, as well as our ability to effectively evaluate the passerine fossil record.
Ongoing palaeontological work in Australia and New Zealand has provided important insight into the origins of several crown passerine lineages (Boles, 1995;Nguyen, 2016Nguyen, , 2019Nguyen et al., 2014Nguyen et al., , 2016Worthy et al., 2010), but Paleogene crown passerine fossils from outside Australasia have thus far proven to be more challenging to interpret. As such, with a few notable exceptions (e.g., Bocheński et al., 2018Bocheński et al., , 2021Manegold, 2008aManegold, , 2008b, the passerine fossil record of Europe has contributed little to our knowledge of the evolutionary origins of most major crown passerine subclades, despite the relative richness of European passerine fossils, some of which represent the oldest known crown passerines (Ballmann, 1969;Manegold, 2008b;Mayr, 2009). In particular, fossil passerines from the Oligocene of Europe (Bocheński et al., 2021;Manegold, 2008b;Mayr & Manegold, 2004, 2006aRiamon et al., 2020) remain enigmatic. Specimens hypothesised to represent two of the deepest passerine subclades, total-clade Passeri ("oscines"; Manegold, 2008b) and total-clade Tyranni ("suboscines"; Bocheński et al., 2021), are known from the European Oligocene (Table 1), yet the precise phylogenetic position of these fossils continues to be controversial. For instance, Wieslochia weissi (Mayr & Manegold, 2004) has been alternatively proposed to represent a stem eupasseran (the most exclusive clade composed of Passeri + Tyranni; Mayr & Manegold, 2006b), a stem group representative of Tyranni , and a crown group representative of Eurylaimidae (broadbills, a subclade of Tyranni; Fjeldså et al., 2020). In light of this ongoing phylogenetic uncertainty, reassessing these specimens in additional detail has a potential to significantly clarify a number of outstanding macroevolutionary questions, such as the biogeographic origins of major passerine subclades. For example, Tyranni is entirely absent from the Palearctic in the present day, but firmly establishing the phylogenetic affinities of these Oligocene passerine fossils could support a European origin for the clade.
Several morphological synapomorphies have been proposed to diagnose the three deepest extant passerine subclades-Acanthisitti, Passeri and Tyranni (Manegold, 2008b;Mayr & Manegold, 2006b;Worthy et al., 2010). However, most major subclades within these groups remain uncharacterised by discrete synapomorphies or diagnostic character combinations, and few detailed anatomical character-taxon matrices exist for crown Passeriformes. A frequent misconception that may partly account for this dearth of research attention is the notion that passerine skeletons are generally 'uniform' (Mayr, 2009(Mayr, , 2013. However, a greater impediment to research may simply be the sheer taxonomic diversity of Passeriformes combined with uncertainty regarding the higher-order phylogenetic relationships of passerines that have only recently been resolved (Harvey et al., 2020;Moyle et al., 2016;Oliveros et al., 2019). The enormous diversity of extant passerines presents a legitimate logistical challenge that has dissuaded attempts to generate large-scale morphological datasets for crown Passeriformes.
Here, we broach this gap in our understanding of passerine comparative morphology by focusing on the carpometacarpus. As a comparatively robust component of the wing skeleton (Figure 1), carpometacarpi exhibit reasonable preservation potential in the fossil record (Ballmann, 1969;Manegold, 2008b;Nguyen et al., 2016), and the impressive morphological variability of the carpometacarpus may have phylogenetic relevance. Although carpometacarpi have provided phylogenetically-informative morphological TA B L E 1 Oligocene passerine fossils included in the phylogenetic analyses.  Manegold (2008b) characters in previous investigations of fossil passerines, available character data for passerine carpometacarpi tend to be either diagnostic only for the deepest clades within crown Passeriformes (Mayr & Manegold, 2006a, 2006bManegold, 2008b;Worthy et al., 2010), or for a small number of family-level passerine subclades Tomek & Bocheński, 2000). In most cases, the phylogenetic information extracted from isolated fossil passerine carpometacarpi has been limited to the ordinal or subordinal level by a lack of adequate comparative data from across passerine diversity. An in-depth analysis of morphological variation of the passerine carpometacarpus therefore has potential to inform patterns of morphological evolution and trait acquisition across passerine phylogeny.
We generated a character matrix of 54 morphological characters of the passerine carpometacarpus (Figures 2-6  Poecile atricapillus FMNH 504323 6(2) 22(1) also included 11 fossil carpometacarpi, see Table 1. These Oligocene fossils were selected based on their morphology being clearly and adequately documented in published literature with good quality photos available for character scoring, and our interest in characterising anatomical transitions relevant to the deepest splits within Passeriformes. We included as many of the earliest crown passerine fossils as possible (early Oligocene in age) that fit the criteria outlined above, as well as several isolated late Oligocene carpometacarpi that are morphologically distinct. We regret that we could not include several specimens of significant interest (described by Bocheński et al., 2011Bocheński et al., , 2013Bocheński et al., , 2018Bocheński et al., , 2021 based on their carpometacarpi being poorly preserved or not clearly discernible in published images. Character optimisation was carried out under parsimony using the Willi Hennig Society version of TNT (Goloboff et al., 2008) (Mayr, 2008) was included in our molecular scaffold as the sister taxon to crown Passeriformes, based on its phylogenetic position and supporting morphological synapomorphies described by Mayr (2008).
Several characters with more than two states were treated as ordered in cases where it was logical to assume that anatomical transitions were likely to pass through separately identified intermediate states (see Table S2). Scripts for all analyses are in Supporting Information.
We conducted four Bayesian phylogenetic analyses to evaluate the affinities of fossil carpometacarpi that were carried out in MrBayes V3.2.2 (Ronquist et al., 2012) using the CIPRES Science Gateway (Miller et al., 2010). We excluded characters 1-5 which are continuous and therefore could not be included. All MrBayes analyses were performed for two runs with four chains and 30 million replicates sampled every 4000 generations with a chain temperature of 0.1 and burn-in set to 0.25. The likelihood model priors were sampled from a gamma distribution with four rate categories with variable coding for morphological data. Analyses 1-4 were summarised as 50% majority rule trees (contype = Halfcompat). The molecular scaffold (see above) was applied to all extant taxa, and Zygodactylus luberonensis was also topologically constrained as the sister group to crown Passeriformes. MrBayes constraint scripts were generated using the package Paleotree (Bapst, 2012) in R V4.1.1 (R Core Team, 2021).

| DISCUSS ION
This study represents the most comprehensive exploration of passerine carpometacarpal morphology to date, with an unprecedentedly diverse taxon sample exceeding that of any previous morphological investigation of passerines. We identified diagnostic combinations of carpometacarpal character states for major passerine subclades, and our phylogenetic analyses recovered fossil passerine carpometacarpi from the Oligocene of Europe as early representatives of major subclades within crown Passeriformes.
Our detailed evaluation of carpometacarpal characters reveals substantial homoplasy in the passerine carpometacarpus, and highlights the need to re-evaluate previously described anatomical characters and incorporate both phylogenetic analyses and character state optimisation to develop informative lists of diagnostic character combinations for key passerine clades. We believe this approach will be necessary for robustly placing fragmentary passerine fossils into an evolutionary context. Nonetheless, these results have implications for re-calibrating the timeline of passerine evolutionary history, and bear on the complex biogeographic origins of one of the deepest passerine subclades, the suboscines (Tyranni).

| Phylogenetic affinities of Oligocene suboscine fossils from Europe
We analysed several controversial crown passerine fossils from the Oligocene of Europe (Figure 7), which have considerable potential to illuminate aspects of early passeriform evolution. Most of these fossils were incorporated into phylogenetic analyses for the first time here. Our phylogenetic results recovered four fossil specimens (SMF Av 504; NT-LBR-014; SMNK-PAL 3980; SMNS 59466/1) as representatives of crown Tyranni across all analyses (crown-group suboscines; Figure 8). These fossils were previously attributed to total-clade Tyranni, but the basis for our assignment differs from prior work (Table 1). Earlier studies suggested that these fossils exhibited diagnostic character states for Tyranni and lacked diagnostic features of Passeri (Manegold, 2008b;Mayr & Manegold, 2006a, 2006b. From our character optimisation results, we find that char. 38(0) (a protruding dorsal surface of the caudodistal region of metacarpal II; Figure 3; Manegold, 2008b) is ubiquitously present across Tyranni, although this character state is plesiomorphic for crown Passeriformes. We find that the alternate state ( as seen in all other crown passeriforms (Mayr & Manegold, 2006b).
However, we argue that the caudodistal portion of metacarpal III is in the same position as seen in most extant suboscine taxa, such as tyrannidans and eurylaimidans, and that its more ventrally oriented craniodistal portion may be hidden, which may also obscure whether the craniodistal portion of metacarpal III extends further distally than the caudodistal portion (

| Phylogenetic affinities of Oligocene oscine fossils from Europe
The remaining crown passerine fossils investigated here were recovered within crown Passeri with strong support (Figure 8), including two carpometacarpi (SMNS 59466/14 & SMF Av 517; Figure 7j,k) that were assigned to Passeriformes incertae sedis (Manegold, 2008b). These two carpometacarpi were originally unassigned to a more specific phylogenetic placement within Passeriformes because they exhibit an unfused intermetacarpal process (char. 29[0]; Figure 4). An intermetacarpal process fused to metacarpal III has long been considered a synapomorphy of crown Passeriformes (Mayr & Manegold, 2006b), and all extant passerines in our sample exhibited this character state (29[1]). It is possible that the unfused intermetacarpal process on these fossils is indicative of a skeletally immature ontogenetic state (Manegold, 2008b), although this remains challenging to assess.   Figure 6), was first mentioned by Manegold (2008b) and used by Fjeldså et al. (2020) Figure 8). The carpometacarpus may exhibit even more homoplasy across Passeri than Tyranni, exemplified by poor phylogenetic resolution within Passeri (see Figures S5 and S6), rendering taxonomic assignment of these fossils particularly challenging.

| Implications for the evolutionary history of crown Passeriformes
We have shown that the passerine carpometacarpus is characterised by widespread homoplasy, epitomised by the lack of truly unique synapomorphies diagnosing passerine subclades. Nevertheless, our detailed evaluation of morphological variation across a broad taxon sample identified various phylogenetically-informative traits. These place constraints on interpretations of the phylogenetic placement of passerine fossils from the Oligocene of Europe. Our phylogenetic analyses corroborate prior hypotheses that representatives of crown Tyranni were present in Europe from the early Oligocene (e.g., Riamon et al., 2020), and co-occurred with representatives of crown Passeri during the late Oligocene (Manegold, 2008b). These results are consistent with the hypothesis that representatives of crown Passeri dispersed out of Australasia by the end of the Oligocene (Moyle et al., 2016;Oliveros et al., 2019).
The oscine fossils sampled in this study are more morphologically disparate than the suboscine fossils studied (Figure 7), and were recovered in several different positions across the Passeri crown group ( Figure 8). By contrast, early Oligocene suboscine carpometacarpi were generally morphologically similar and were recovered as one another's respective sister taxa (Supporting Information), while the late Oligocene suboscine fossil examined here diverged somewhat in its morphology. All the suboscine fossils sampled most likely derive from different species based on size differences (Manegold, 2008b;Mayr & Manegold, 2006a, 2006bRiamon et al., 2020). Amongst the early Oligocene suboscine fossils, our phylogenetic analyses indicate that Wieslochia weissi is neither a stem-group eupasseran (Mayr & Manegold, 2006b) nor a crown eurylaimid (Fjeldså et al., 2020) as previously hypothesised, but likely resides somewhere within crown Tyranni, as opposed to falling within its stem group . Further investigation across the skeleton of articulated specimens will be necessary to clarify their phylogenetic placement within total-clade Tyranni. From our analyses and evaluation of diagnostic character states, it seems plausible that the early Oligocene suboscine specimens investigated here are closely related to oneanother and potentially represent stem-group members of either Tyrannides or Tyrannida.
Delineating between plesiomorphic and apomorphic character states has frequently proven challenging due to high rates of homoplasy across the passerine carpometacarpus, complicating the assessment of the phylogenetic position of fossils. The work presented here has exemplified the necessity to study passerine comparative morphology in a broadly-sampled phylogenetic context and has demonstrated that at least one character-rich element of the skeleton-the carpometacarpus-exhibits significant morphological lability. Characterising additional elements of the passerine skeleton at a comparable level of detail will be critical for furthering our understanding of passerine morphological evolution, and clarifying key patterns from the earliest stages of passerine evolutionary history.

AUTH O R CO NTR I B UTI O N S
EMS and DJF conceptualised the project. All authors contributed to data collection. EMS and JMTN compiled the character list. EMS

O PE N R E S E A RCH BA D G E S
This article has earned Open Data badge. Data and materials are available at www.morph osour ce.org/proje cts/00045 4063/about ?local e=en.

DATA AVA I L A B I L I T Y S TAT E M E N T
All 3D models are available from Morphosource and links to individual media can be found in Table S1. All other datafiles are provided in the Supporting Information.