Not Enough Skeletons in the Closet: Collections-Based Anatomical Research in an Age of Conservation Conscience

Abstract

The emergence of new technologies and improved computing power helped to introduce a renewed vitality in morphological research in recent decades. This is especially apparent in the new advances made in understanding the evolutionary morphology of the skeletal system in extinct and extant squamate reptiles. The new data generated as a result of the recent increase in attention are relevant not only for systematic analyses but also are valuable in their own right for contributing to holistic perspectives on organismal evolution, mosaic evolution in the rates of change in different anatomical systems, and broader patterns of macroevolution. A global community of morphological researchers now can share data through online digital collections, but opportunities for continued advance are hindered because we lack even basic data on patterns of variation of the skeletal system for virtually all squamate lineages. Most work on skeletal morphology of squamates is based on a sample size of n = 1; this is an especially noticeable phenomenon for studies relying on X-ray computed tomography technology. We need new collections of skeletal specimens, both material and digital, and new approaches to the study of skeletal morphology. Promising areas for continued research include the recent focus on skeletal elements not traditionally included in morphological studies (especially systematic analyses based upon morphological data) and efforts to elucidate patterns of variation and phylogenetically informative features of disarticulated skeletal elements. Anat Rec, 297:344–348, 2014. © 2014 Wiley Periodicals, Inc.

A new age of discovery is under way in the study of skeletal morphology of squamate reptiles and the application of attendant data to evolutionary questions. This intellectual renaissance was shaped by several breakthroughs and opportunities that emerged over the last two decades. The most prominent of those breakthroughs were the emergence of new technologies for visualizing skeletal morphology, the development of rapid computing power and new systematic methods that permit novel treatment of traditional morphological data for systematic analysis, and an as-yet sluggish but increasingly prominent attention to detailed reinterpretation and documentation of the fossil record. Indeed, the only sector of the field that recently seemed locked in an anachronistic past is the material component of that research program—the specimen collections on which we base our studies. Even here, real change is afoot, as digital collections of morphological data gain both in prominence and in data content. There is a real need for online databases that document phenotypic data; excellent examples are DigiMorph (http://www.digimorph.org/), Morphobank (http://morphobank.org/), and the nascent Morphometry.org (http://morphometry.org/). These are digitally based collections that are served online and that seek to link a global community of researchers around a shared body of morphological imagery and interpretation. However, because they are largely based on material specimens housed in traditional institutional collections they continue to suffer from many of the same limitations that plague our material collections. The most prominent and disturbing of those limitations is one of sample size; as a community, squamate morphologists seem stuck in an almost pre-Darwinian and pre-Wallacian world view in which relatively little attention is paid to the patterns of variation expressed in the skeletal system.

OBSTACLES TO UNDERSTANDING VARIATION

Documenting patterns of skeletal variation within and across taxa remains an important goal for those interested in evolutionary morphology. Yet, despite widespread acceptance of the role of variation in shaping evolutionary dynamics, detailed studies that document patterns of skeletal variation are available for surprisingly few taxa of squamate reptiles. Given the ascendancy of molecular biology and the shifting hiring priorities at many academic institutions, it is perhaps not surprising that vertebrate paleontologists increasingly (but not exclusively) assumed responsibility over the last few decades for the development of datasets that specifically acknowledge and embrace skeletal variation as a biological reality, and a topic of interest in itself, not merely a sideshow or an impediment to the adequate elucidation of evolutionary relationships (Abdala, 2006). Because paleontology is necessarily a serendipitous effort, the extant biota provides the best and most ready opportunities to explore patterns of skeletal variation within target groups of interest.

Integration of detailed fossil records with adequate voucher-based collections of extant taxa has great potential to elucidate simultaneously patterns of evolutionary morphology in both spatial and temporal dimensions. However, efforts to understand evolutionary morphology of squamate reptiles still are hampered by challenges posed by established operational and cultural practices within both the paleontological and the herpetological communities.

From the paleontological side, those challenges include a long-standing tradition of examining and publishing descriptions of only a limited number of selected skeletal elements and operational practices in the identification of fossils that go significantly beyond the observed morphology of the specimens themselves. Those traditions are most obvious in the voluminous literature documenting the fossil record of Neogene and Quaternary squamates, the majority of which consists of disarticulated and disassociated skeletal elements. There is an undeniable bias in the skeletal elements that are reported in the paleontological literature for different squamate groups. Smaller skeletal elements of all groups may simply be lost during collection because they pass through screens with mesh sizes insufficiently small to retain them (e.g., Mead et al., 1982, 1992). That collections bias may be especially important for collections made prior to the development of modern screening techniques and would apply uniformly to all small skeletal elements, regardless of the taxon from which they were derived. But elemental bias in the literature goes beyond that. It is clear, for example, that the literature on fossil snakes is dominated by descriptions of vertebral elements. Similarly, the published literature for lizards is replete with descriptions of the maxilla, dentary, frontal, and parietal elements of the skull, with fewer (but still notable) descriptions of the premaxilla and quadrate. That pattern also is reflected in museum collections, in many of which the lizard fossils that are catalogued from unpublished localities consist only of those same elements. Certainly, some elements of the postcranial skeleton are readily preserved (e.g., vertebrae and some limb bones) but are notoriously difficult to identify even to more inclusive groupings of lizards. It is certainly true that cranial elements of lizards are considered to be the most important for taxonomic identification and for systematic analysis. The reasons for the bias among cranial elements are not entirely clear. In some cases a preservational bias may be responsible; the elements listed above are among the more durable elements of the skull. But the ossified braincase elements (especially when they are co-ossified), pterygoids, prefrontals, and jugals of many lizards also are durable and are readily preserved. In developing, reviewing, and using fossil collections of squamate reptiles over the last several decades, we noted that most cranial elements of lizards, even relatively delicate ones, are preserved as fossils, but generally are unrecognized and can be found in “scrap,” “unidentifiable,” or “junk” bone containers.

We suspect, therefore, that the element bias derives from the historical practices associated with the development of those collections (noted below) and from a lack of familiarity with the skeletal system of lizards among most paleontologists. Most of the collections of Neogene and Quaternary fossils were made by paleontologists who were seeking (and were most familiar with) mammalian fossils. Their perspective on the relative merits of isolated skeletal elements was, therefore, shaped by the traditions of mammalian paleontology, a specialization under which tooth-bearing elements were and are held to be of pre-eminent value for fine-scale taxonomic resolution. It would be a logical step for those trained under such a tradition to conclude that the dentigerous elements of lizards must hold equal value, thus explaining the importance initially placed on the premaxilla, maxilla, and dentary. The frontal and parietal would follow as relatively large and recognizable elements that can readily be seen to vary interspecifically. There is no doubt, however, that meaningful phylogenetic signal can be preserved in other skeletal elements, and that those elements can, consequently, be identified to less inclusive taxonomic levels. When all preserved skeletal elements are studied and described, a more holistic and meaningful perspective can be gained of the fossil record and of the evolutionary and morphological history of lineages. Recently published works demonstrate that in at least some cases skeletal morphology of isolated elements alone can be sufficient for achieving fine-scale taxonomic and phylogenetic resolution. Species-level resolution and a fully resolved phylogeny were obtained by including character data from “non-traditional” disarticulated cranial elements (Bhullar, 2011), and reconstructions of detailed cranial anatomy of extinct lizards were made possible with careful attention to even fragmentary remains of all cranial elements (Smith, 2009, 2011).

An important additional bias impacting the interpretation of especially Neogene and Quaternary squamate fossils is that identification of specimens often is facilitated with reference to the modern geographic distributions of extant taxa. In many cases, modern biogeography serves only as a means of restricting the pool of species with which comparisons are made when identifying fossils. In other cases, modern geographic distributions are ultimately the basis of the taxonomic identification of a fossil. Both practices introduce bias into the interpretation of fossils, a problem exacerbated by the fact that those practices often are not stated explicitly, and are, as a result, “covert” and less than obvious to other scientists seeking to use paleontological data in other contexts (e.g., for conservation biology, large-scale studies of paleobiogeography, or calculating estimates of the timing of lineage divergence). Those problems were discussed at length by Bell et al. (2010), but the important point here is that those methodological approaches can mask interesting data about mosaic evolution in rates of morphological change because they promote a false sense of the reliability of certain elements to yield fine-scale taxonomic resolution, and they can hide patterns of homoplasy that, as a result, remain unexplored.

An improved understanding of evolutionary morphology of squamates also is hindered by biases generated from within the herpetological community. That community operates under long-established norms that govern voucher-based collections and, thus, dictate the nature of materials that are readily available for morphological study. In most collections around the world, the standard preservation for squamate specimens is alcoholic, often of formalin-fixed specimens, and skeletal preparations are viewed by many modern biologists as “wasted specimens” that are to some extent unworthy of the space dedicated to house them. When skeletal specimens are preserved they often are specimens garnered from the pet trade or from field-based collections from which data were accidentally lost. Regardless of source, they often lack data on sex, ontogenetic age, body mass, locality, date of collection, and even basic measurements. Without such basic data, those specimens may be useful for addressing some questions but are almost useless for exploring detailed questions about evolutionary morphology. Of equal importance now is that many skeletal preparations are not accompanied by skins or by tissue samples. Numerous taxonomic revisions were made in recent decades and were based on either detailed scalation characters or molecular characters, or both. Taxonomic identifications of skeletal specimens that lack locality data, tissues, or skins cannot easily be updated, so that many historical skeletal collections are, in many cases, taxonomically antiquated as well.

A NEED FOR NEW COLLECTIONS

Evolutionary morphologists require adequate voucher-based collections in the same way that other subdisciplines of biology do. The attitude that skeletal specimens are “wasted” specimens needs to be confronted head-on. As does the prevailing view that a collection may be considered adequate when it contains a single specimen of each of the taxa from the geographic mandate established by a given institution; such “stamp collections” are adequate for quick surveys of morphological disparity, but are not helpful for most modern questions about morphology, especially those dealing with the patterns, sources, and drivers of morphological variation.

It is true that in many respects existing collections are inadequate for many questions in biological science. Today the need for adequate collections of such material is ever more acute. Moreover, the nature of the material needed is changing. Historically, skeletal preparations were held to be valuable only if the elements (of the skull at least) remained articulated. The morphological data that were used in many phylogenetic analyses of squamate relationships were based predominantly on articulated skulls and skeletons, and the majority of characters were based on, or could be scored on, articulated specimens (Estes et al., 1988; Lee, 1998, 2005; Caldwell, 1999; Conrad, 2008; Gauthier et al., 2012). It is likely that those traditional preparations (e.g., articulated skulls) have yielded the majority of the data they will yield, but the recent demonstrations of the utility of disarticulated elements for phylogenetic reconstruction and for facilitating identification of fossils (Bhullar, 2011; Smith, 2009, 2011) demonstrate a need for collections that include disarticulated material as well. Such collections are extraordinarily rare and were universally built by herpetologists with strong interests in paleontology (collections at The University of Florida, The University of Texas at Austin, Michigan State University, East Tennessee State University, and Georgia College stand as rare exemplars of excellent collections of disarticulated skeletons of squamates).

Technological advances such as high-resolution X-ray computed tomography (CT) offer a partial (albeit biased) solution to the paucity of material. Advances in CT technology, including planar area detectors with sufficient resiliency and speed to hold up under bombardment, enabled a dramatic decline in the costs associated with CT scanning in the last 15 years. For example, at The University of Texas CT facility in Austin, the hourly charge for CT scanning in 1997 was $104 USD/hr, and a scan of a moderately large lizard skull took perhaps 13 hr, with a resultant 5 µm resolution. Today, the cost is $108 USD/hr (the hourly rate pays the time for trained personnel, and the change over 15 years is shockingly low), and a comparable specimen can be scanned in 36 min with a resolution of 0.2 µm. Nonetheless, CT scanning remains a somewhat expensive means of acquiring morphological data, and many authors continue to rely on single exemplars when generating CT datasets. A notable example is the massive recent contribution to squamate morphology published by Gauthier et al. (2012). The total number of specimens examined for that work was an unparalleled 1,319 yet the CT data that were the basis of the evaluation of skull morphology (Gauthier et al., 2012:7) were generated for only a single exemplar of each species (Maisano, personal communication, May, 2013). Expense may be a partial explanation for the fact that for the vast majority of published CT-based morphological studies of squamates the sample size included was n = 1. The n = 1 phenomenon is noticeable in historical studies based upon CT data, with early departures being those by Bever et al. (2005), Boughner et al. (2007), and Nance (2007), each with a sample size of only two specimens. Some more recent authors used greater sample sizes, exemplified by studies published by Banzato et al. (2012; n = 6, 4, and 3, respectively for three taxa), Costantini et al. (2010; n = 6 each for three taxa), Polachowski and Werneburg (2013; n = 14), and Starck et al. (2012; n = 11). But it is not the expense of CT scanning alone that causes the n = 1 phenomenon, because there is a real and general paucity of data on skeletal variation for squamates, and only in relatively few cases did sample sizes exceed 1 in modern studies (e.g., Rieppel and Crumly, 1997; Barahona and Barbadillo, 1997, 1998; Barahona et al., 1998; Bell et al., 2003, 2009; Conrad, 2004; Daza et al., 2008, 2009; Olori and Bell, 2012).

The time and need to gather, evaluate, and report data on patterns of variation across broad sample sizes is now certainly upon us. An n = 1 approach relegates perspectives on variation to a subsidiary status under an antiquated philosophy in which single specimens are taken as representative exemplars of entire lineages. As a community of scholars, squamate morphologists should know that this certainly is not the case. We need new collections of skeletal materials that include samples from across the geographic range of species, that encompass both sexes, and that include specimens from a diversity of ontogenetic ages. Building and housing such collections, whether material or digital, is not a trivial exercise and will require a community of scholars dedicated to the task. Simultaneously, museums and other academic institutions also will need to make the investment in hardware and logistical support to store, archive, and (we hope) make available digital datasets of morphological data.

CONCLUSION

The reticence to prepare new skeletal specimens that can be used to generate the requisite data noted above clearly has cultural, institutional, financial, and inertial underpinnings. We encourage our colleagues to renew and intensify efforts to build collections that will serve as the foundation for those requisite data, but we realize that we are advocating for voucher-based collections in a cultural and political climate in which increasingly stringent restrictions govern voucher-based collections. This is exacerbated for endangered, rare, or poorly known species, whose conservation status stands as a strong disincentive to voucher-based collections (e.g., Roberts and Solow, 2008; LaDuc and Bell, 2010). The consequences of diminished voucher-based collections are nontrivial, but the development of such collections poses novel challenges in the modern world (for recent discussions see, e.g., Hoberg et al., 2009; Boakes et al., 2010; LaDuc and Bell, 2010; Winker et al., 2010; Rivers et al., 2011; Casas-Marce et al., 2012), and such collections are required for significant advances to be made in understanding patterns of evolutionary morphology within and across squamate lineages. We remain convinced that those collections can be made in a way that ensures a maximum data yield from every specimen taken. For example, it is a relatively simple thing to record basic mensurative data and body mass for every specimen taken. Sex data usually can be recorded for all but the youngest individuals. Stomach contents can be easily retained (as can field notes and museum records indicating empty stomachs), and data on orientation of prey items can be recorded. Field notes also can be used easily to record basic ecological data, including vegetation associations, behavioral observations at or just before capture, and so on. All specimens can be tissued immediately, and representative skins can be retained and preserved as dry or alcoholic specimens. Tissue samples serve both molecular and morphological communities because retention of tissues not only makes the specimen useful for subsequent molecular-based studies but also guarantees the continued importance and utility of the skeletal specimen in the face on ongoing taxonomic revisions that increasingly are based exclusively (or predominantly) upon molecular data.

Morphological research is far from moribund, and many exciting opportunities present themselves now for integrating paleontological and neontological perspectives to develop broad-scale appreciation of the evolutionary dynamics that shaped the squamate tree of life. Our current collections, and the traditional philosophies and practices that shaped their development and growth, clearly are inadequate and will not permit the morphological community to capitalize on those opportunities. We require more and different skeletons in our closets!

ACKNOWLEDGEMENTS

The authors thank Rich Ketcham for data on historical trends in costs associated with CT scanning and Jessie Maisano for information on trends in use of CT technology for studies of squamate morphology. Juan D. Daza and Dennis Trombatore provided important assistance with literature. Our thoughts on this topic were shaped by conversations with many colleagues over the last two decades. Significant among those are Aaron Bauer, Jason Anderson, Gabe Bever, Anjan Bhullar, Jim Bowden, Jack Conrad, Nick Czaplewski, Juan D. Daza, Paul Doughty, Susan Evans, Jacques Gauthier, Harry Greene, Jason Head, Marci Hollenshead, Ric How, Howard Hutchison, Marc Jones, Walter Joyce, Maureen Kearney, Alicia Kennedy, Travis LaDuc, Patrick Lewis, Jessie Maisano, Brad Maryan, Lyn Murray, Jen Olori, Ted Papenfuss, Jim Parham, Dennis Parmley, Blaine Schubert, Krister Smith, Dave Steadman, Geraldine Swartz, Jens Vindum, and David Wake.