Portal de Periódicos da CAPES

Undisciplined thinking: morphology and Hennig’s unfinished revolution

2008; Wiley; Volume: 33; Issue: 1 Linguagem: Inglês

10.1111/j.1365-3113.2007.00411.x

1365-3113

Quentin D. Wheeler,

Mycorrhizal Fungi and Plant Interactions

There was a time, not long ago and prior to Hennig (1966), when taxonomy was widely dismissed as a mere service to 'real'– read experimental– sciences. Taxonomists were regarded by many to have nothing more to contribute to modern biology than the pragmatic role of identifying species and keeping track of their names. This was a legacy of the conflation of systematics with genetics by Huxley (1940), Mayr (1942) and others (see Wheeler, 1995, 2008a). Hennig re-elevated taxonomy, as phylogenetic systematics, to its rightful place as a rigorous, free-standing and central field of the biological sciences. Taxonomy is typically performed best when it is carried out for its own sake. Taxonomists are motivated to explore species, character diversity and phylogenetic relationships within monophyletic groups. The ultimate goal of taxonomists is a phylogenetic classification with associated scientific names, what Hennig described as biology's general reference system. Oh yes, they make species identifiable, too. Current molecular initiatives, including DNA barcoding and DNA taxonomy, threaten to reduce species discovery as well as classifications to nothing more than a service. Because 'new' species would be 'discovered' on the basis of phenetic distances only, DNA barcoding might be described better as a disservice to biology (Prendini, 2005; Wheeler, 2005). After all, it offers only arbitrary averages, by contrast with explicitly testable alternatives, such as the phylogenetic species concept (Wheeler & Platnick, 2000). DNA taxonomy (in the sense of Tautz et al., 2003) is another flawed approach that would diminish the information content of classifications (e.g. Lipscomb et al., 2003). The trend in molecular phylogenetics ('phylogenetic biology') has been to increasingly marginalize the evidential basis of taxonomy and to treat the creation of 'trees' largely as a service to those same 'real' sciences. It is not coincidental that central to taxonomy 'done for its own sake' is a detailed study of characters, whereas most molecular phylogenies are published without characters. Morphology is regarded increasingly as an optional adjunct to molecular phylogeny, rather than an essential component of our understanding of species and their diversification. It is said that morphology provides few characters by comparison with molecular sources, and that its complex characters are inconveniently difficult to interpret and unacceptably subjective in their analysis. As a result, we are educating few young comparative morphologists and homogenizing 'phylogenetic biology' to be effectively little more than molecular 'trees' and morphology little more than literature reviews. The number of projects in which serious original morphological study is undertaken is alarmingly few, meaning that existing morphological hypotheses are untested, mistaken homology assessments are perpetuated and new morphological characters are undiscovered. The tragic loss represented by this neglect of comparative morphology has yet to be felt. Molecular phylogenies have concerned themselves largely with the reinterpretation of known taxa, and especially those bearing controversial 'key' morphological achievements, such as the evolution of flight in insects, vascular tissues in plants and, of course, the great tritomy of human, chimp and gorilla. Increasingly, it seems that molecular phylogenies are performed in service to others, to provide chronological patterns to biologists studying some attribute of organisms of sufficient complexity to be of inherent interest. The same putative virtue of simplicity that is claimed to make the interpretation of phylogeny from molecules less subjective deprives amino acids of sufficient complexity to store historical information. That only emerges from the next step in complexity: the sequence of amino acids in strands of DNA. Just as this minimal complexity adds some imperfect historical information, increasingly complex characters – genes, functionally linked genes, proteins, folding patterns, tissues, cells, ontogenetic pathways and morphology – each in turn potentially stores more and more historical information. This information is more difficult to interpret. Nevertheless, as has been well established, it can be made fully explicit and testable (Hennig, 1966; Farris, 1979; Gaffney, 1979; Nelson & Platnick, 1981; Wiley, 1981; Schuh, 2000). It takes time to begin to interpret the techniques used by Leonardo da Vinci to capture the mystique of Genevre D'Benci, and considerably less time to see the minimal technique behind Robert Rauschenberg's White Painting. Are we to believe that it is not worth the investment of time in the case of the former to appreciate such content, or that all art should be monochromatic like the latter in order to make visits to art museums more time efficient? Could it be that it is complexity itself, including the seemingly endless variety in morphology, that draws us to study evolution and its history, and that having cladograms divorced from credible morphology and other characters is a vacuous endeavour, except as a service to others who do have 'interesting' character sets? Is it possible that there is a good reason that the same people who continue to visit zoos to marvel at giraffe necks, zebra stripes and rhinoceros horns would probably not make a trip to a public place to admire the percentage of differences between fragments of cytochrome oxidase subunit I genes amongst a comparable number of faceless samples? Molecular phylogenies have interest only by virtue of what they can tell us of the evolution of complex morphological characters, other characters or spatial distributions. The reality is that the external morphology of the vast majority of known species is described very incompletely, particularly in mega-diverse taxa such as insects. Moreover, the majority of characters and homology assumptions in groups such as insects are too infrequently tested to be considered to be corroborated. Furthermore, because an estimated 75% of all living species of insects remain unknown to science (Grimaldi & Engel, 2005), the morphology of the majority of species is entirely undocumented. Considering the imminent threats of extinction and the many species that are the last vestiges of nearly extinct clades, this would seem to be a particularly inopportune moment to abandon morphology. Those who advocate classifications that no longer require the preservation of whole specimens in collections (e.g. A. P. Vogler, personal communication), or allow species to be described only for which sequences are at least potentially available (Dayrat, 2005), are terribly misguided (Valdecasas et al., 2007) and ride on the tide of political correctness in support of all things molecular. This has all the intellectual depth of an advertisement for improved laundry detergent. Sadly, it sells. What we need in actuality is not DNA-exclusive or DNA-intensive classifications, but integrative taxonomy (e.g. Will et al., 2005). We need a taxonomy (a general reference system, a phylogenetic classification, including associated Linnaean names) that is informed by and summarizes as many sources of evidence as possible. Hennig (1966) referred to this as 'holomorphology'. Combining excellent analyses of morphology, fossils, ontogeny and molecules will advance systematics much further than molecular phylogenies alone or in combination with superficial treatments of morphology or fossils or ontogeny. We need a taxonomy that integrates the expertise, diligence, knowledge, passion and talents of teams of specialists, each contributing something unique and excellent to collaborative studies of taxa. Considering that as many as three million insect species await discovery and that hundreds of thousands may not survive the present century, there would seem to be enough work for an army of full-time specialists of every stripe. If someone is a good collector and that is what he or she enjoys and excels at, then let us provide support for collecting; our collections are growing at an embarrassing rate in the midst of the biodiversity crisis, and we can afford to encourage and support as many collectors as can contribute to the effort. If someone is interested in comparative morphology, then let us support him or her to be the best morphologist he or she can be, and not require a dilution of good morphology by prescribed molecular laboratory work. If another is interested in molecular techniques, then let us sustain that interest, and so forth. There are valuable roles for specimen preparators, illustrators, phylogenetic analysts, palaeoentomologists, ontogeneticists, comparative ethologists, taxonomic cyberinfrastructure engineers, and on and on. Let us celebrate the diversity of interests and motivations that draw a diversity of people into systematic entomology, and not attempt to define what a good 'systematist' is so narrowly as to at once drive many talented contributors from the effort and homogenize and stifle the science to the point that it loses its most remarkable character: its innate ability to integrate many efforts towards a common goal and diverse sources of data into a reliable general reference system. The current fashion of multidisciplinary science is good in principle, but unfortunately has been implemented in such a way that taxonomy is expected to conform to assumptions and quantitative methods that are appropriate instead to experimental sciences. It seems to me that, in general, this is achieved more easily through transdisciplinary teams rather than multidisciplinary individuals. Into these teams the community of amateur taxonomists should be integrated, although their numbers appear also to be declining (Hopkins & Freckleton, 2002). Hennig's (1966) concept of 'holomorphology' celebrated the long-established practice of taxonomy as a field capable of synthesizing evidence from all relevant comparative sources (e.g. Simpson, 1961). Hennig taught us that what matters is not the source of data, but rather how the data are analysed and whether special similarities are differentiated from raw overall similarity. The potent implications of Hennig's theories have yet to be fully realized. Rather than welcoming the input of credible data from all sources, we have allowed technology and political correctness to drive the field towards one source of data and away from analytical methods that are consistent with Hennig's theory of special similarity. Both are to the detriment of taxonomy. Real taxonomy is and has always been focused as much upon character analysis and evolution as upon the discovery of species and reconstruction of relationships. With the grand challenge questions of taxonomy unanswered (Cracraft, 2002; Page et al., 2005), we need to return to the core strengths of taxonomy, including explicit studies of characters. For the purpose of cladistic analysis, we simply need data that can and are analysed appropriately to reveal cladistic (as opposed to phenetic) patterns. In the context of cladistic analysis, molecules and morphology are just data, nothing more, nothing less. We need the best data we can have, and should not seek to homogenize the activities of those who gather it. Whether limited by talent, motivation, time or resources, many of us would prefer to focus on where our passions lie in order to make inspired contributions of the highest quality possible. For some of us, that means focusing exclusively on morphology or molecules or ontogeny or fossils. What matters is that as much data as possible are available; let us not confuse the need for taxonomy to have multiple data sources with an expectation that the full range is generated by each taxonomist. The discipline of taxonomy stands apart from the many fields of functional or general biology (Nelson & Platnick, 1981). It has taken centuries to refine the theoretical and epistemological justifications and assumptions of taxonomy; these advances should not be sacrificed on the altar of political correctness and modernity, in order to receive acceptance from experimental or molecular biologists. Students can be trained in a matter of months to sequence DNA proficiently and, if they generate large enough sums of external funding, can become celebrated as 'good' systematists. Morphologists must be educated over a period of years to work to high levels of excellence; they must master, in many cases, centuries of existing knowledge, a vast and technical vocabulary of terms required to discuss precisely complex characters and their component parts, understand intimately the theory behind concepts such as homology and synapomorphy as well as their use in a particular taxon, demonstrate a critical 'eye' capable of recognizing significant patterns of similarities or differences, learn to use alternative data sources as independent tests of sameness, and know how to collect, prepare, dissect and illustrate or image specific parts in a way that records and communicates complicated bits of visual knowledge. I do not suggest that there are not excellent molecular systematists that go far, far beyond the rote generation and analysis of data caricatured above; there are, and they deserve the same respect and support as good morphologists. The problem today, however, is not that they are denied stature and resources, it is that they alone receive them within systematic biology at the great detriment of the general reference system and growth of knowledge of morphology in particular. What is interesting at the heart of taxonomy is understanding the patterns of origin and diversification of complexity, not merely enumerating 'species' about which nothing is known concerning their structural uniqueness or relationships. Unless we support adequately the many facets of taxonomy, we impoverish our science to the point of undermining its importance. More importantly, unless we recognize that taxonomy is non-experimental (it cannot be commingled with, for example, ecology or population genetics) and that it is not subject to the same statistical tools appropriate to tokogenetic systems (Wheeler, 1995, 2004, 2008c), we necessarily diminish its unique contributions to biology. Combining experimental and historical sciences can only result in a kind of undisciplined thinking that ignores the epistemic foundations of each discipline. The quality of taxonomy depends upon the recognition of and respect for the differences between taxonomy and experimental biology. The diverse skill sets mentioned above (from collectors to morphologists to molecular systematists) could, if coordinated as an effective team, realize incredible gains in the advance of taxonomy without sacrifice of the diversity and quality of data. I propose that we assemble some model teams with such diverse talents in defiance of the conformity that has been imposed upon the field by an overindulgence of the demands of molecular techniques. The notion that molecular data are 'new' is of course deeply flawed. Molecular sequences are the organic codes for the many morphological characters we have not yet discovered. The genes and morphological characters are, necessarily, of precisely the same evolutionary age. If the molecules do not impart the modernity that diverts funds into molecular work, what does? Obviously, it is the technology. This is indeed impressive. Anyone doubting that need only compare what we could do before and after polymerase chain reaction techniques were introduced. Quietly, however, similar advances are taking place in technologies used to study, document, analyse, visualize and communicate findings in comparative morphology. Digital microscopy, computer-assisted tomography, MorphBank and MorphoBank, visualization tools, image recognition and a host of other advances are developing equally impressively. If morphology can be approached with comparably new technology, has more information content and its hypotheses are subject to rigorous testing, there would seem to be no legitimate reasons for its continued neglect. National Science Foundation (NSF) director, Arden L. Bement, has captured the excitement surrounding the next generation of cyberinfrastructure: 'At the very heart of the cyberinfrastructure vision are cultural communities that support peer-to-peer collaboration and new modes of education. They are distributed-knowledge communities in an institutional context, not of bricks and mortar like the traditional university, but rather virtual organizations that work across institutional boundaries – and ultimately around the globe'. This reveals the potential for taxonomic research when distributed natural history collections, instruments, research resources and specialists within and between taxa are seamlessly linked to create, test, improve and share all available knowledge of a taxon (Wheeler, 2008c). The LINNE workshops have forged a vision for cybertaxonomy: the fusion of integrative descriptive taxonomy with computer science and engineering (Page et al., 2005; Wheeler, 2008b). Armed with such infrastructure, taxonomists would at last be poised to address taxonomy's 'big questions' (Cracraft, 2002; Page et al., 2005) on an appropriately planetary scale. To realize the unprecedented opportunities of a domain-specific cyberinfrastructure for taxonomy, we must address the full range of needs of taxonomy from the molecular to the morphological and from the growth of collections to the engineering of new tools. The power of international teams and cooperation amongst museums is evident in the Planetary Biodiversity Inventory awards from the NSF (Knapp, 2008; Page, 2008). As impressive as these projects are, they are limited by existing taxonomic tools. The kind of research environment, distributed instrumentation and 'species observatory' capability envisioned in the LINNE cyberinfrastructure would drastically alter the speed and quality of such collaborative work (Page et al., 2005; Wheeler, 2008b). Some look starry eyed into a future with handheld devices that sequence a bit of tissue and spit out instant species identifications (Janzen, 2004). These are, by and large, users of taxonomy, not taxonomists; those motivated to identify a specimen, not understand a species in the context of its evolutionary history. Students of mathematics can use calculators to obtain answers to problems without bothering to learn the principles of mathematics but, in so doing, they deprive themselves of the education they need. In a similar way, identifying species in utter ignorance of anything about them, most especially their morphology and habits, can provide answers to questions, but reduce taxonomy to a service rather than a science, or even a serious past time. The Achilles' heel of these 'barcoders' is the theoretical weakness of DNA barcoding as it is currently advocated (Prendini, 2005; Wheeler, 2005; Little & Stevenson, 2007). It can be a tremendously useful and exciting new tool in our arsenal of species identification methods. However, we have only two choices in its implementation. Either we support taxonomists to discover and describe species that are subsequently identified by the use of DNA barcodes, or we reduce species from rigorously testable hypotheses to arbitrary phenetic clusters. Technical expediency is a poor bargain when divorced from theoretical rigour. If the epistemology of taxonomy is ignored, taxonomy itself will wither. It is time to undertake unapologetic studies of morphology, employing powerful new digital tools. It does not take much vision, courage or leadership today to follow the now well-worn path of molecular phylogenetics. I challenge a few brave young scientists passionate about comparative morphology and willing to undertake a career as a new breed of morphologist to do so, and decline the artificial constraints imposed by political correctness. Together with computer scientists and engineers, who can build whatever tools imaginative morphologists can dream of, they could contribute to international teams focused on rapid progress in our knowledge of species, phylogeny and classification. Morphology is only important if you want to understand the evolution of species on this planet. We can create our own fortunes, as have countless generations before. There are many good reasons to perform and support comparative morphology studies beyond making complex, information-rich characters available; I mention only a few. Morphology frequently allows field biologists to observe plants and animals in the field without the need for destructive sampling. Morphological characters are the only bridge between fossil and Recent species. Morphological characters are frequently the object of natural selection. The intellectual challenges and rewards associated with deep thinking about homology and apomorphy are simply without compare. In spite of proponents of the PhyloCode, who dismiss the only evidence that exists for either species or monophyletic groups in an effort to make a questionable philosophical point about the individuality thesis, it is characters and their analysis that are at the heart of taxonomy. The work of morphology is far from complete. It is important to recall that our knowledge of morphology and anatomy is incomplete for all species, fragmentary for most species and non-existent for the majority of living species that are unknown to science. The level of detail of morphological description for most insects is superficial. Moreover, it remains that every homology statement is an hypothesis in need of repeated critical testing, as are statements about the distributions of characters amongst species (Nelson & Platnick, 1981). Unless we continue to discover, clarify, corroborate and map distributions of morphological characters, much of the rationale for doing phylogenies – the need to interpret special similarities in their historical context – is gone. I agree with those who suggest that the Hennigian revolution is unfinished (e.g. Nelson, 2004). Neglect of morphology has stunted our understanding of the most interesting aspects of evolutionary history. A disproportionate focus on molecular phylogenies has contributed to a widening disconnection between phylogenetic analyses and formal classifications and names (e.g. Franz, 2005). Contributing factors to this disassociation include global parsimony at the neglect of individual character analysis, a focus on tree construction as opposed to the quality of data entered in matrices, and the abundance of molecular data that have made morphological datasets appear anaemic by comparison. The theoretical advances associated with a clear distinction between pattern and process, that distanced phylogenetic theory from the subjectivity of the evolutionary taxonomy of preceding decades, has seemingly been forgotten (Eldredge & Cracraft, 1980; Nelson & Platnick, 1981). In its place is a newfound confusion of pattern with process, as in maximum likelihood assumptions, hailed as progress (Nei & Kumar, 2000). As mentioned above, it is reasonable to describe DNA barcoding and much of the analysis of molecular data as phenetic and a return to yet other assumptions debunked early in the Hennigian revolution. Molecular data are analysed as mere similarity, ignoring a central tenet of Hennig's theory. The extent to which this is an acceptable assumption rests in large measure on the extent to which mutations are clock-like, a questionable assumption (Schwartz & Maresca, 2006). As my colleague Dr Antonio Valdecasas recently reminded me, molecular and morphological datasets are only partially independent. The code and the coded for are inextricably linked, but in currently unspecifiable ways. Total evidence analyses mix these potentially dependent characters with the hope that they add only noise to a clear phylogenetic signal. When we fail to engage in comparative morphology studies in a rigorous way, we also 'dumb down' morphological characters by failing to capitalize on their information-rich complexity. It is a 'common claim from molecular systematists (is) that a large number of independent characters are needed to be able to estimate phylogenetic relationships robustly and reliably (Rokas et al., 2003)' (Wahlberg et al., 2005). That the number of molecular datasets available today is limited for most taxa (Wahlberg et al., 2005) is problematic in this regard. Moreover, even if whole genomes were widely sequenced for representatively large numbers of infraspecific specimens, it would remain that popular neo-phenetic methods of molecular analysis would rest upon unproven assumptions about molecular clocks (Schwartz & Maresca, 2006). This adds, I think, considerable impetus to take morphology seriously, analysing morphological characters carefully to ensure the benefits of their strong historical signal. A signal that is so strong that it can sometimes be detected even when combined with a comparative sea of simple molecular data: 'Even though all characters are given equal weight and the morphological characters are vastly outnumbered by the molecular data, the low intrinsic homoplasy of the morphological features allows them to establish a structural framework in the combined analysis to which the morphological data contributes synergistically' (Wahlberg et al., 2005, p. 1583). It is critical that we act immediately to educate the next generation of morphologists, before much hard-earned knowledge of morphology is lost to science. It is not enough for the science of taxonomy that we have phylogenies generated from molecular evidence and summaries of past morphology studies. New morphological characters must be explored and documented and existing ones revisited and tested, as exemplified by the revisions of reduviid bugs (Weirauch, 2008) and xeromelissine bees (Packer, 2008) later in this issue. The NSF's Partnerships for Enhancing Expertise in Taxonomy (PEET) (Rodman & Cody, 2003) provided an impetus for the transfer of taxonomic, including morphological, expertise between generations, and indeed products of the scheme appear in this issue (Hardy et al., 2008; Unruh & Gullan, 2008). Nonetheless, many of the best morphologists have retired or are approaching retirement and there seems to be little effort to replace them. Moreover, the vast majority of morphological hypotheses are but rarely tested. I strongly urge that, as taxonomy progresses into its next stage, what we might call the 'New Taxonomy' (Wheeler, 2008c) or 'Cyber-Enabled Taxonomy', that we use the emerging cyber-infrastructure to re-establish vibrant supplies of diverse data, especially morphology. Every taxon should have a 'knowledge community' of experts distributed around the world collaborating to contribute to the growth of knowledge of its species. Although taxonomists who wish to study more than one source of data should be free to do so, that should not be a mandate. It is through the full-bore, take-no-hostage approach to each discipline that the very best science will emerge. Taxonomy should be a transdisciplinary science, drawing the best from a wide range of researchers, not a concatenation of superficially practised techniques by undisciplined investigators. I thank Dr Peter S. Cranston for the opportunity to contribute this essay, and Dr Frank T. Krell (Denver Museum of Nature and Science, Denver, Colorado, U.S.A.), Dr Kelly B. Miller (University of New Mexico, Albuquerque, New Mexico, U.S.A.), Dr Antonio G. Valdecasas (Museo National Ciencias Nagurales, Madrid, Spain) and Dr David M. Williams (The Natural History Museum, London, U.K.) for insightful comments.