[Palaeontology, 2020, pp. 1–16] CATEGORICAL VERSUS GEOMETRIC MORPHOMETRIC APPROACHES TO CHARACTERIZING THE EVOLUTION OF MORPHOLOGICAL DISPARITY IN OSTEOSTRACI (VERTEBRATA, STEM GNATHOSTOMATA) by HUMBERTO G. FERRON 1,2* , JENNY M. GREENWOOD1, BRADLEY DELINE3,CARLOSMARTINEZ-PEREZ 1,2,HECTOR BOTELLA2, ROBERT S. SANSOM4,MARCELLORUTA5 and PHILIP C. J. DONOGHUE1,* 1School of Earth Sciences, University of Bristol, Life Sciences Building, Tyndall Avenue, Bristol, BS8 1TQ, UK; [email protected], [email protected], [email protected] 2Institut Cavanilles de Biodiversitat i Biologia Evolutiva, Universitat de Valencia, C/ Catedratic Jose Beltran Martınez 2, 46980, Paterna, Valencia, Spain; [email protected], [email protected] 3Department of Geosciences, University of West Georgia, Carrollton, GA 30118, USA; [email protected] 4School of Earth & Environmental Sciences, University of Manchester, Manchester, M13 9PT, UK; [email protected] 5School of Life Sciences, University of Lincoln, Riseholme Hall, Lincoln, LN2 2LG, UK; [email protected] *Corresponding authors Typescript received 2 October 2019; accepted in revised form 27 February 2020 Abstract: Morphological variation (disparity) is almost aspects of morphology. Phylomorphospaces reveal conver- invariably characterized by two non-mutually exclusive gence towards a generalized ‘horseshoe’-shaped cranial mor- approaches: (1) quantitatively, through geometric morpho- phology and two strong trends involving major groups of metrics; and (2) in terms of discrete, ‘cladistic’, or categorical osteostracans (benneviaspidids and thyestiids), which proba- characters. Uncertainty over the comparability of these bly reflect adaptations to different lifestyles. Temporal pat- approaches diminishes the potential to obtain nomothetic terns of disparity obtained from categorical and insights into the evolution of morphological disparity and the morphometric approaches appear congruent, however, dis- few benchmarking studies conducted so far show contrasting parity maxima occur at different times in the evolutionary results. Here, we apply both approaches to characterizing history of the group. The results of our analyses indicate that morphology in the stem-gnathostome clade Osteostraci in categorical and continuous data sets may characterize differ- order to assess congruence between these alternative methods ent patterns of morphological disparity and that discrepancies as well as to explore the evolutionary patterns of the group in could reflect preservational limitations of morphometric data terms of temporal disparity and the influence of phylogenetic and differences in the potential of each data type for charac- relationships and habitat on morphospace occupation. Our terizing more or less inclusive aspects of overall phenotype. results suggest that both approaches yield similar results in morphospace occupation and clustering, but also some differ- Key words: disparity, morphospace, categorical data, geo- ences indicating that these metrics may capture different metric morphometrics, Osteostraci. Q UANTIFICATION of morphology and morphological (Webster & Hughes 1999; Barisß 2017), taxonomic or disparity is integral to assessing general macroevolution- geographic scale (Butler et al. 2012; Deline et al. 2012), ary patterns in the fossil record, such as adaptive radia- environmental distribution (Hopkins 2014), community tions, rates of evolution, responses to extinctions, biotic structure (Deline 2009), sampling of phenotypic charac- replacements and the existence of constraints on form ters (Foth et al. 2012; Hopkins 2017), choice of metric (Foote 1997 and references therein). Consequently, there (Ciampaglio et al. 2001), character selection (Deline & have been many attempts to evaluate the impact of Ausich 2017) and choice of methodology (Villier & potential biasing factors on morphological disparity esti- Eble 2004; Hetherington et al. 2015; Hopkins 2017; mates, including the possible influence of taphonomy Romano et al. 2017). © The Palaeontological Association doi: 10.1111/pala.12482 1 2 PALAEONTOLOGY Morphological disparity and morphospace occupation vertebrates. However, cranial shape characters contribute patterns have conventionally been based on two non- extensively to the systematics of the group (Sansom mutually exclusive approaches to characterizing morphol- 2009a), making osteostracans an ideal focus for bench- ogy. Firstly, through geometric morphometrics, morphol- marking the characterization of morphology for disparity ogy is characterized quantitatively in terms of continuous analysis using discrete vs categorical characters. We char- variation in aspects of organismal shape and proportions acterize the morphology of this clade based on the dis- (e.g. Stayton & Ruta 2006). This approach is most com- tinctive and character-rich headshield, both in terms of monly used to assess morphospace occupation at lower geometric morphometric and discrete categorical ‘cladis- taxonomic levels due to the need for homologous features tic’ data. We compare perceptions of morphological dis- in all specimens within a single study (Wills et al. 1994; parity based on these data sets and, further, interpret the Bookstein 1997). Therefore, while these methods are very results in terms of their implications for understanding sensitive to morphological variation, their effectiveness in this temporal, phylogenetic and ecological context of the characterizing overall phenotype diminishes with increas- evolution of this key clade. ing taxonomic scale reflecting the concomitant reduction in the number of universally shared homologous features (Hetherington et al. 2015). Alternatively, morphology can Osteostraci as a case study be characterized in terms of discrete or categorical charac- ters of the sort most commonly employed in the cladistic Osteostracans constitute a clade of extinct jawless verte- analysis of morphology. This approach to characterizing brates that range from the Llandovery (lower Silurian) morphology is usually undertaken in analyses of broad to the Frasnian (Upper Devonian) (Sansom 2009a). taxonomic scope, where there are large numbers of cate- Osteostracans are a taxonomically and morphologically gorical differences between taxa, but fewer universal diverse group, characterized by a semicircular dermal homologous structures that might serve as a basis for headshield encompassing the cranial and pectoral geometric morphometrics. As such, the categorical char- regions, a number of cephalic fields, pineal foramen, acterization of morphology can capture more unique nasohypophysial opening, and a postcranial body cov- aspects of form (Briggs et al. 1992; Wills 1998). However, ered in thick scales (Janvier 1996). Cornual and/or ros- this approach is also much less sensitive to finer changes tral processes extending from the headshield are well in shape and proportion. Despite the differences between developed in a number of independent lineages and these approaches, most of the small number of bench- paired fins are present in many taxa but lost in others marking studies that have been conducted have shown (Janvier 1985; Sansom 2008), or they evolved conver- that continuous and categorical approaches to character- gently between jawed vertebrates and derived Osteostraci izing morphology yield broadly congruent patterns of dis- (Denison 1951). Most osteostracans have headshields parity (Villier & Eble 2004; Anderson & Friedman 2012; that are strongly oblate dorsoventrally although some Foth et al. 2012; Hetherington et al. 2015; Hopkins 2017; have a more approximately circular or prolate in cross- Romano et al. 2017; Schaeffer et al. 2019). Nevertheless, sectional profile (Janvier 1996) (Fig. 1A–B). Remarkably, Mongiardino Koch et al. (2017) have shown that these despite being a potential source of valuable biological two approaches yield contrasting patterns of disparity in and ecological information (e.g. Janvier & Lawson 1985; their analysis of the scorpion genus Brachistosternus. This Belles-Isles 1987; Bunker & Machin 1991; Afanassieva difference may reflect the relative power of continuous 1992; Mark-Kurik 1992; Janvier 1996; Morrissey et al. versus categorical approaches to characterizing morphol- 2004; Davies 2009), such morphological disparity has ogy at different taxonomic scales. However, this study not been analysed quantitatively. The headshield is the may also reflect the fact that benchmarking studies aspect of osteostracan anatomy that varies most through remain small in number and many more such studies are osteostracan phylogeny, as well as being the anatomical needed to obtain nomothetic insights into whether differ- division most commonly preserved. Hence, it is the ent approaches to characterizing morphology impact on source of almost all characters that have been exploited the perception of the ensuing patterns of morphological in osteostracan phylogenetics (e.g. Janvier 1985; San- disparity and the evolutionary processes that are inferred som 2009a). Geometric morphometric characterization from them. To that end, we explore the evolution of of the osteostracan headshield will, therefore, provide morphological disparity in osteostracans, the extinct clade for a fair and effective comparison to existing categori- of jawless stem-gnathostomes that is generally perceived cal characterizations of osteostracan morphological varia- to be most closely related to jawed vertebrates (Janvier tion, facilitating insights