The closest evolutionary relatives of pterosaurs: What the morphospace occupation of different skeletal regions tell us about lagerpetids

Abstract

Exquisite discoveries and new interpretations regarding an enigmatic group of cursorial avemetatarsalians led to a new phylogenetic hypothesis regarding pterosaur affinities. Previously thought to be dinosaur precursors, lagerpetids are now considered to be the closest relatives to pterosaurs. This new hypothesis sheds light on a new explorable field, especially regarding the character acquisition and evolution within the pterosaur lineage. In the present study, the morphospace occupation of distinct skeletal regions of lagerpetids within the morphological spectrum of avemetatarsalians is investigated. This approach indicates which portions of the skeleton are more similar to the anatomy of pterosaurs and which portions present different homoplastic signals. The analyses demonstrate that the craniomandibular traits of lagerpetids are pterosaur-like, the pectoral girdle and forelimb are dinosauromorph-like and the axial skeleton and the pelvic girdle and hindlimb are unique and highly specialized among the analyzed sample. So, despite the close phylogenetic relationships, the postcranial skeleton of lagerpetids and pterosaurs are very different. The occurrence of two distinct and highly specialized groups of pterosauromorphs coexisting with a wide ecological range of dinosauromorphs during the Triassic suggests pressure for new niches occupation.

1 INTRODUCTION

The first vertebrates to evolve active flight, pterosaurs have been studied during years (Witton, 2013). The anatomy, biomechanics, biology, and evolution of the group were exhaustively investigated (e.g., Aires et al., 2021; Kellner, 2003; Palmer & Dyke, 2010; Yang et al., 2019). Nevertheless, its origins and early evolution are poorly known (Baron, 2021; Ezcurra et al., 2020). Whereas the identity of the sister group of Pterosauria was a mystery for centuries, lagerpetids were thought to be dinosaur precursors, or nondinosauriform dinosauromorphs (Baron et al., 2017; Cabreira et al., 2016; Garcia et al., 2019; Irmis et al., 2007; Müller et al., 2018; Nesbitt, 2011; Sereno & Arcucci, 1993). Characterized by elongate and gracile hindlimbs, much of the skeletal anatomy of lagerpetids remains unknown (Langer et al., 2013; Müller et al., 2018). However, our knowledge of the osteology of this group was recently improved by new discoveries and interpretations of additional bony elements (Baron, 2021; Cabreira et al., 2016; Ezcurra et al., 2020; Kammerer et al., 2020; Martz & Small, 2019; McCabe & Nesbitt, 2021). These new data supplied a new interpretation regarding the phylogenetic affinities of lagerpetids and pterosaurs (Baron, 2021; Ezcurra et al., 2020; Kammerer et al., 2020). Instead of dinosaur precursors, the new hypothesis places these animals as the sister group of pterosaurs. This is particularly interesting because the origin of pterosaurs is obscure and poorly understood (Sereno, 1991. Benton, 1999; Ezcurra et al., 2020; Baron, 2021). The oldest unequivocal pterosaurs are Norian in age (Dalla Vecchia, 2013; Wild, 1978). Nevertheless, there are few well-preserved specimens of early pterosaurs (Dalla Vecchia, 2013) and the pterosaur affinities of some of them have been questioned (Soares et al., 2013). The group has been treated as prolacertiforms (Peters, 2000), early archosauriforms (Bennett, 1996; Benton, 1985) or ornithodirans (Benton, 1990; Nesbitt, 2011; Novas, 1996). Therefore, the new hypothesis (i.e., Ezcurra et al., 2020) provides insights on the possible ancestral anatomy of pterosaurs and their closest relatives.

In the present study, I reconstruct the morphospace occupation of distinct skeletal portions of avemetatarsalians to investigate the distribution of lagerpetids regarding the morphospace area of pterosaurs. This analytical approach allows me to observe which portions of the skeleton are more similar to the anatomy of pterosaurs and which portions present different homoplastic signals (e.g., convergence with other avemetatarsalians; Novas et al., 2015). Moreover, this method helps to identify highly specialized regions of the skeleton.

2 MATERIALS AND METHODS

A series of morphological disparity analyses were performed to explore the morphological occupation of different skeletal regions of lagerpetids among the avemetatarsalians morphospace. The dataset for the present analyses was derived from the cladistic dataset of Ezcurra et al. (2020), which includes 823 characters. It has been demonstrated that cladistic data summarize morphological variation in comparable ways to geometric morphometric approaches (Hetherington, 2015). Similarly, some authors found that discrete and continuous character-based disparity analyses provide similar results (Romano et al., 2017). The former dataset was divided into nine subsets (see Appendix S1): 823 total characters; 428 craniomandibular characters; 390 postcranial characters; 350 cranial characters (excluding lower jaw and dental characters); 54 lower jaw characters (excluding dental characters); 24 dental characters; 101 axial skeleton characters; 93 pectoral girdle and fore limb characters; and 181 pelvic girdle and hindlimb characters. Taxon sampling was reduced to include aphanosaurs, dinosauromorphs, pterosaurs, and lagerpetids only. Other groups were not included because the aim of this investigation is to explore the homoplastic signals within Avemetatarsalia. Lehmann et al. (2019) demonstrated that higher percentages of missing data biases the results of disparity analysis. Therefore, operational taxonomic units (OTUs) with excessive missing entries (i.e., “?”) were pruned from the subsets (see Appendix S1). A Euclidean distance matrix (EDMA) was calculated from each subset using the software MATRIX (Wills, 1998). During this step, the discrete traits were transformed in continuous values employing a Generalized Euclidean Distance according to Wills (1998, 2001):

$${\mathrm{GED}}_{\mathit{ij}}=\sqrt{{{\sum }_k{W}_{\mathit{ijk}}{S}_{\mathit{ijk}}}^2}$$

where GED_ij is the distance between taxa i and j, W_ijk is an integer, depicting the character weight, and S_ijk is the linear Euclidean character distance (Wills, 2001). In addition, weighted mean values are assigned when distances cannot be calculated because of missing entries (“?”). In this case, the weighted mean values are given based on the calculable S_ijk values (Wills, 2001). For further information regarding this step see Wills (2001). Then, a principal coordinate analysis (PCo) was performed for each EDMA with the multivariate package GINKGO (Bouxin, 2005). Following similar studies (Butler et al., 2012; Pacheco et al., 2019), the centroid of all OTUs was taken as the origin of multivariate axes, and the Cailliez method of negative eigenvalue correction was adopted. The PCo 1 and 2 of each analysis (see Appendix S1) were used to construct the bivariate graphs using the software PAST (Hammer et al., 2001).

3 RESULTS

The PCos 1 account from 6.85 to 22.73% of variance, whereas the PCos 2 account from 6.31 to 20.3% of variance (Figure 1a–i; see Appendix S1 for complete list of variances). In the first analysis, which employs all characters, most lagerpetids occur on an exclusive quadrant (Figure 1a). However, there is a lagerpetid (i.e., Kongonaphon kely) that plots within the convex hull of pterosaurs. Kongonaphon is next to the pterosaur Raeticodactylus. Both share an elongated anterior ramus of the maxilla and the ascending process of the maxilla forming an oblique angle with the main axis of the bone (Kammerer et al., 2020; Stecher, 2008). Except for that lagerpetid, no other taxa occur on the common area of pterosaurs, which is distinct from that of dinosauromorphs and aphanosaurs. In the PCo 1, aphanosaurs, and dinosauromorphs overlap lagerpetids, whereas in PCo 2 all of the groups overlap with pterosaurs. In the analysis employing craniomandibular characters, lagerpetids occur on the same quadrant of some pterosaurs and are distinct from the morphospace of other groups (Figure 1b). Lagerpetids overlap pterosaurs in the PCo 1. Conversely, both groups do not overlap in the PCo 2. Employing postcranial characters, lagerpetids occupy an exclusive area outside any other convex hull (Figure 1c). When the axes are observed separately, there is a small overlap between lagerpetids and pterosaus. In the PCo 1, lagerpetids overlap aphanosaurs and dinosauromorphs, whereas it does not occur in the PCo 2. Lagerpetids are in the same quadrants of pterosaurs in the analysis that employs cranial (without lower jaw and dental traits) characters (Figure 1d). Both groups are separate from aphanosaurs and dinosauromorphs in the PCo 1. In the analysis of lower jaw characters, lagerpetids occur within the morphospace of pterosaurs (Figure 1e). In this analysis, there is some degree of overlap among the morphospaces of pterosaurs and dinosauromorphs. The pterosaur Preondactylus bufarinii is within the area of dinosauromorphs. This is interesting, because this taxon lacks the edentulous and tapering anterior end of the dentary (Dalla Vecchia, 1997), which is present in some early pterosaurs (Ezcurra et al., 2020). The PCo 1 segregates lagerpetids and dinosauromorphs. In the analysis of dental characters, there is some degree of overlap among the morphospaces of pterosaurs and dinosauromorphs; however, lagerpetids occur outside any convex hull (Figure 1f). The PCo 1 fails to segregate lagerpetids from pterosaurs and dinosauromorphs. On the other hand, lagerpetids are separated from other groups in the PCo 2. Regarding the analysis derived from axial characters, lagerpetids are in the same quadrant as aphanosaurs, which is well-separated from the morphospace occupied by pterosaurs and dinosauromorphs (Figure 1g). Lagerpetids are separated from pterosaurs in the PCo 1 and 2. Nevertheless, lagerpetids overlap aphanosaurs and dinosauromorphs in the PCo 1. Lagerpetids occupy a region occupied by dinosauromorphs when the analysis is run only with pectoral girdle and fore limb characters (Figure 1h). Ixalerpeton plots within the morphospace of dinosauromorphs. Pterosaurs are separated from other groups along the PCo 1, whereas the group overlap lagerpetids and dinosauromorphs in the PCo 2. In the analysis employing pelvic girdle and hindlimb characters, lagerpetids occupy an exclusive area outside the convex hull of other groups (Figure 1i). Lagerpetids overlap pterosaurs in the PCo 1. In the PCo 2, both groups are separated.

FIGURE 1

Bivariate plots showing the results of the morphospace occupation analyses. (a) Using all the characters. (b) Using craniomandibular characters. (c) Using postcranial characters. (d) Using cranial characters. (e) Using lower jaw characters. (f) Using dental characters. (g) Using axial characters. (h) Using pectoral girdle and forelimb characters. (i) Using pelvic girdle and hindlimb characters. (j) Reconstructed skeleton of a lagerpetid (by Maurício S. Garcia) summarizing the results of the analyses. Orange reversed triangles correspond to aphanosaurs, blue circles corresponds to dinosauromorphs, purple triangles correspond to pterosaurs, and red “X” corresponds to lagerpetids. Silhouettes based on the artwork by Márcio L. Castro, Gabriel Lio and Corey Ford

4 DISCUSSION AND FINAL REMARKS

The analyses demonstrate that most of the shared morphology between lagerpetids and pterosaurs relies on the cranial and mandibular regions (Figure 1d, e). Indeed, both clades share several neuroanatomical (e.g., subtriangular and dorsoventrally tall floccular fossa of the braincase) and mandibular traits (e.g., edentulous anterior end of the dentary that tapers to a point; lack of splenial; Ezcurra et al., 2020). This provides insights on the possible ancestral anatomy of Pterosauria as hypothesized by phylogenetic bracketing and homology. The dental traits of lagerpetids occupy an exclusive area on the morphospace of avemetatarsalians; however, lagerpetids plot close to the convex hull of pterosaurs (Figure 1f). This may reflect some peculiar traits present in both clades, such as the absence of dental serrations and presence of cusps (Ezcurra et al., 2020). The occurrence of comparable mandibular and dental traits favors similar feeding habits (Barrett, 2000; Cabreira et al., 2016; Stocker et al., 2016). Nevertheless, the absence of overlapping areas of morphospace indicates distinct degrees of dental specializations.

The postcranial anatomy of pterosaurs is extremely specialized (e.g., vertebrae fused to form the notarium; fore limbs modified as wings; extremely hollow bones; Nesbitt, 2011; Kellner, 2003; Aires et al., 2021; Baron, 2021) and distinct from that of lagerpetids, which is clear in the analysis employing pectoral and fore limb characters (Figure 1h). The axial skeleton of lagerpetids is similarly peculiar, occupying a unique area in the morphospace (Figure 1g). Indeed, the vertebrae of lagerpetids are clearly distinct from that of other avemetatarsalians (Sereno & Arcucci, 1993), lacking laminae radiating from the apophyses, pneumatic features, and accessory joints (Cabreira et al., 2016). Similarly, the sacrum is composed solely by two vertebrae (Cabreira et al., 2016; Sereno & Arcucci, 1993), retaining the plesiomorphic condition among archosaurs (Nesbitt, 2011). Despite the presence of features shared between pterosaurs and lagerpetids regarding the pelvic girdle and hindlimb (e.g., strongly ventrally extended pubo-ischiadic plate; hook-shaped proximal head of the femur; co-ossified astragalus and calcaneum; Ezcurra et al., 2020; Baron, 2021), both groups occupy an exclusive area on the morphospace when these traits are analyzed (Figure 1i). This reinforces the peculiar and unique limb anatomy of lagerpetids, which is usually associated with cursorial habits (Arcucci, 1989; Kammerer et al., 2020; Nesbitt et al., 2017). Aphanosaurs and dinosauromorphs also occupy an exclusive area. This is possibly associated with the locomotor diversity experienced by the distinct clades of avemetatarsalians (Pintore et al., 2021), especially if the hindlimb was attached to the wing in pterosaurs via a patagium (Bennett, 2015; Elgin et al., 2011; Palmer, 2018; Witton, 2008). Furthermore, Sereno and Arcucci (1993) pointed to a set of pelvic girdle and hindlimb traits in Lagerpeton chanarensis that suggest it may have been saltatorial. Some of these traits include a small pelvic girdle, fused proximal tarsals, and a very narrow didactyl pes. Curiously, Sereno and Arcucci (1993) recognizes that pterosaurs have a comparably small pelvic girdle.

In sum, the craniomandibular traits of lagerpetids favors pterosaur affinities, the pectoral girdle, and forelimb are dinosauromorph-like and the axial skeleton and the pelvic girdle and hindlimb are unique and specialized (Figure 1j). Nevertheless, there are distinct approaches designed to investigate the morphospace occupation of extinct animals employing discrete traits (e.g., Apaldetti et al., 2021; Brusatte et al., 2008; Lloyd, 2016; Novas et al., 2015). These distinct methods may produce different results, so additional investigations on the morphospace occupation of lagerpetids and pterosauromorphs are welcome.

Despite the close phylogenetic relationships (Baron, 2021; Ezcurra et al., 2020; Kammerer et al., 2020), the postcranial skeleton of lagerpetids and pterosaurs displays highly distinct morphologies. This reflects the distinct behaviors of both groups. It is interesting because the occurrence of two distinct and highly specialized groups of pterosauromorphs coexisting with a wide ecological range of dinosauromorphs during the Late Triassic suggests pressure for new niches occupation. Whereas dinosaur and dinosaur relatives are widespread and adapted to terrestrial ecosystems (Langer et al., 2013; Marsh & Parker, 2020; Marsola et al., 2019; Michael et al., 2018), pterosaurs evolved flight and lagerpetids may have been able to climb (Ezcurra et al., 2020). So, despite the distinct postcranial anatomy, the similar craniomandibular (including neuroanatomical) features led both groups to explore two distinct niches. A certain ecological diversity also appears to occurred within Lagerpetidae. Some members of the clade (e.g., Kongonaphon kely from Middle to late Triassic of Madagascar) are extremely miniaturized, with a femoral length about to 40 mm (Kammerer et al., 2020), whereas some lagerpetids (e.g., NMMNH P-18091 from the early Late Triassic of North America) achieved a femoral length approximately of 200 mm (Beyl et al., 2020). This size disparity may reflect some degree of distinct ecological roles (Woodward et al., 2005). For instance, the feeding habits of lagerpetids are poorly understood (e.g., Cabreira et al., 2016; Kammerer et al., 2020); however, even if Kongonaphon kely and NMMNH P-18091 were faunivorous, both probably preyed on distinct prey types, demanded different energetic values and were preyed upon by distinct predators (Owen-Smith & Mills, 2008; Woodward et al., 2005). In addition, animals from the same lineage with distinct body sizes usually evolve adaptations related to body mass increase (e.g., Rauhut et al., 2011; Tsai et al., 2020). Therefore, there is a hidden and wide spectrum of ecological and morphological variation to uncover regarding these pterosaur relatives.

ACKNOWLEDGMENTS

The author thanks Adam Marsh and Sebastián Rozadilla for their useful comments and suggestions that greatly improved the manuscript. In addition, the author thanks Matthew Wills for the software MATRIX.