Molecular and Classical Taxonomy Workshop
7 – 13 November, 1994
by Douglas Eernisse
André Adoutte (Université Paris XI, France), Fred Breeuwer (University of Amsterdam, The Netherlands), Anne Bruneau (University of Reading, UK), Piero Cammarano (Universita di Roma, Italia), Rüdiger Cerff (Universität Braunschweig, Deutschland), Russel Chapman (Louisiana State University, USA), Rob DeSalle (American Museum of Natural History, New York, USA), Douglas Eernisse (California State University, Fullerton, USA), Patrick Forterre (Université Paris-Sud, France), Mari Källerjö (Swedish Museum of Natural History, Stockholm, Sweden), Costas Krimbas (University of Athens, Greece), Charles Marshall (University of California, Los Angeles, USA), Richard Olmstead (University of Colorado, USA), Jeanine Olsen (Univeristy of Groningen, The Netherlands), Niels Petersen, organizer, (European Science Foundation, Strasbourg, France), Hervé Philippe (Université Paris-Sud, France), Andrew Smith (The Natural History Museum, London, UK), Wytze Stam (University of Groningen, The Netherlands), Simon Tillier (Laboratoire des invertébrés marins et Service de systématique moléculaire, Paris, France), Arndt Von Haeseler (Ludwig Maximilians Universität, München, Deutschland), David Williams, organizer, (The Natural History Museum, London, UK).
Cette rencontre, qui a eu lieu sur proposition de l’European Science Foundation (Strasbourg), visait à explorer les relations entre les systématiques morphologique et moléculaire.
Compte rendu (en anglais)
A workshop on Molecular and Classical Taxonomy was held on 8-13 November 1994 hosted and co-sponsored by Fondation des Treilles in Tourtour, France. This was the second of five workshops planned by the European Science Foundation (ESF) to foster improved communication among the European community of systematic biologists. The goals of the four remaining workshops were established during the initial ESF workshop, held at the Naturel History Museum of London on 14-15 January 1994. It was decided that the workshops should strive to unite European systematists across national boundaries through improved communication, to create a decentralised systematic resource, to define and mobilize Europe’s contribution to tackling the scientific challenges of biodiversity, to explore the relationships between morphological and molecular systematics (the subject of the present workshop), to consider new developments in systematic theory, practice and application, to establish priorities for and consider ways of promoting the training of systematists, and to provide a platform for European cooperation with similar developments in the United States, whose systematists have organized a Systematics Agenda 2000.
Molecular and Classical Taxonomy
Phylogenetic estimates can be applied with great power to many questions of comparative biology. The historical perspective that evolutionary systematics provides is central to all disciplines of biology. Systematists are no longer focused exclusively on “morphological” estimates of phylogeny derived from anatomical, developmental, and paleontological studies, or taxonomy classification. The field of systematics has recently been rejuvenated by an explosion of interest in molecules as a source of phylogenetic data, a contemporary emphasis on quantitative aspects of systematic methodology, and a conceptual shift toward emphasizing strictly phylogenetic groupings in taxonomy classifications. Technological advances such as polymerase chain reaction (PCR) have already led to vast and rapidly expanding gene or protein sequence libraries. Systematists employ comparisons of sequence data for inferring support for genealogical hypotheses. This workshop focused on the interest of many who have recognized that, while molecules hold great promise as a source of data, many problems remain in how these data should be integrated with morphological data sets or phylogenetic estimates.
ESF representative, Niels Petersen, and chairman of the workshop, David Williams (Naturel History Museum, London), organized a diverse program featuring presentations and discussion from 21 European and American systematists in attendance, representing nine countries. The program proceeded from discussions of theory, to more empirical case histories of the phylogenetic analyses of diverse organisms, and concluded with a general discussion of challenges that face the European systematics community.
David Williams introduced the workshop with an overview of the theoretical aspects of obtaining phylogenetic estimates from diverse molecular or morphological sources. The examples chosen by Williams, including published systematic studies involving many of the workshop’s participants, posed the theoretical challenges that successful integration of these data sets will necessarily entail. The workshop presentations began with emphasis on theoretical subjects and continued with empirical case histories. Empirical studies were presented in descending taxonomy generality, beginning with broad considerations of the tree of life, and continuing with increasingly specific comparisons of particular taxonomy groups.
Determining the quality of molecular data
Charles R. Marshall, Mari Källerjö, Richard Olmstead and Hervé Philippe led discussions on efforts to assess the quality of molecular or other data and emphasized the importance of careful data exploration. With the recent advent of PCR techniques, it is now routinely possible to rapidly generate a substantial amount of directly comparable DNA, RNA, or protein sequence data from a variety of organisms of interest, requiring only minute amounts of tissue. Electronic gene and protein databases are consequently expanding at an exponential rate, and much of this data is being analyzed for its systematic content. It is now clear that available data for a particular group of organisms can vary tremendously in how appropriate it is for estimating relationships. Thus, great care is required in the selection of what to sequence and what to include in any subsequent analyses.
Charles R. Marshall led a discussion on methods of tree reconstruction, issues of tree “robustness”, testing for statistical inconsistency, and why congruence between multiple phylogenetic methods may tell us nothing or something, depending on the case. The strengths of DNA or RNA sequence data are that information comes in discrete units, there are an enormous number of potential characters, with a small number of possible character states. Yet one must be able to recognize when a phylogenetic signal is positively misleading. Marshall discussed problematic aspects of prior estimates of amniote phylogeny based on 18S (small subunit) ribosomal RNA (rRNA). Marshall demonstrated that sensitivity to sequence compositional biases could explain why these 18S rRNA results indicated an unexpected bird – mammal grouping, rather than the more conventional, morphological and paleontological based, view of birds as a reptilian lineage, most closely associated with crocodiles.
Mari Källerjö considered alternative methods for demonstrating a statistically significant structure in a data set of molecular or morphological characters. It can be shown that particular, best supported, tree hypotheses differ significantly from trees chosen at random. The hope is that this significant structure may be attributed to an underlying hierarchical pattern of shared ancestry, although none of the methods can guarantee that this is the source of the structure. One method seeks to attribute significant skewness in the distribution of obtained tree lengths to correlation between characters, which may be due to a common ancestry. Källerjö and collaborators have contrived examples that illustrate that other factors, besides correlation among characters, can result in significant skewness. Another method, called permutation (or TPT) testing, compares observed data to randomization of those data. While insensitive to character distributions, a significant result can be obtained even when multiple, mutually incompatible, trees are equally well supported by the data set.
Källerjö prefers a third method, called “total support”, that calculates the total of ail « Bremer Support Index” (also known as “support index” or “SI” or “decay index”) values for a tree, which is meaningful in comparison to values determined for competing trees. This calculation reflects the number of additional steps required to collapse each particular node of the tree when data is analyzed in a parsimony context. A new program called “RNA” from collaborator, S. Farris, features fast total support calculations.
Richard Olmstead considered the distinction between the robustness and accuracy of a phylogenetic result. Robustness pertains to the tendency for a result to remain unchanged when more data of the same kind is collected. Accuracy concerns how perfectly a result reflects actual phylogenetic history. Examples from the already extensive sequence data base of flowering plants (angiosperms) were featured. For example, a recently published data set of over 500 angiosperm rbcL chloroplast DNA sequences was reanalyzed, along with some of the equal number of additional sequences that have subsequently become available. While these sequence comparisons have illuminated relationships within particular angiosperm lineages, relationships among these major lineages remain poorly understood. This may be due to insufficient numbers of informative characters, problems of long branches attracting, or problems of too much evolution on terminal branches that leads to difficulty in reconstructing deeper branching patterns. In contrast, Olmstead argued that ancient hybridization events were unlikely to be responsible for this lack of resolution. Olmstead also gave an example of how combined analysis of different molecular data sets can yield more than the sum of analyses of partitioned data sets. Results from three data sets that were mutually contradictory when analyzed separately were shown to progressively support a combined data result, when pairs of data sets were combined and analyzed.
Hervé Philippe has programmed a JACKMONO method to tease apart the important parameters affecting sequence analysis. He illustrated these methods with various gene sequences from the recent divergence of cetacean mammals (e.g., whales, etc.), the ancient divergence of major eukaryotic lineages, and a variety of simulated data. Performance was judged by how bootstrap proportion or “BP” values were affected as the number of informative sites was increased. Using both parsimony and distance methods, Philippe considered how BP values were affected when various contrasting data types were considered. In particular, Philippe has investigated the signal from transitions versus transversion substitutions, alternative codon position substitutions, sequences with differing G C compositional biases, and analyses with differing taxonomy representation.
Should molecular and morphological data sets be combined?
Systematists are increasingly aware that phylogenetic estimates are dependent on fundamental methodological assumptions, and there is ongoing debate on which assumptions are necessary or supported. For example, a systematist who is interested in obtaining the best genealogical estimate for a group of organisms must decide whether to combine all available data, analyze data sets separately and somehow combine the separate results, or favor particular data set (s) while dismissing others.
Douglas Eernisse discussed competing paradigms of phylogenetic inference, known as “taxonomy congruence” and “total evidence”. In taxonomy congruence, character data are partitioned into sets that are a priori regarded by the investigator as distinct (e.g., morphology and molecules, or gene A and gene B), then one or more tree is derived from analysis of each set, and finally ail resulting trees are somehow combined into a best estimate of phylogeny, most often with a consensus approach. In the total evidence approach, the investigator combines ail relevant character data into a single matrix, and one or more best estimates of phylogeny are derived from this single data set. Eernisse gave two empirical examples of analyses that demonstrated the power of the total evidence approach. In the fore mentioned (Marshall’s) case of incongruent estimates of amniote phylogeny, the conventional view is supported when relevant rRNA, protein, morphological, and paleontological results are considered simultaneously. In contrast, no resolution is apparent in the strict consensus tree of ail the separate analysis results. Another case concerned metazoan (i.e., multicellular animal) phylogeny. Analysis of combined morphological and 18S rRNA data supported an unconventional “Eutrochozoa” grouping of segmented worms (annelids) with mollusks and other minor spiralian phyla, exclusive of arthropods.
David Williams defended a consensus (taxonomy congruence) approach for particular situations, for example, when a single cladistic history cannot be assumed. Williams drew parallels between the analysis of biogeographic history and phylogenetic inference. He further argued that “component” analysis, developed first by Nelson and Platnick for biogeographic studies, has a direct relationship to the character analysis fundamental to phylogenetic inference by representing characters, like taxa, as single hierarchical cladograms. Williams questioned whether characters are directly observable, and suggested that one ultimately is performing a consensus of numerous separate three-taxon statements. According to this view, consensus and total evidence become the same thing when data are appropriately treated.
Jeanine Olsen summarized the preceding theoretical discussions, which mostly involved alternative phylogenetic reconstruction methodologies. It was recognized that additional concerns remain. For example, there has been relatively little attention directed to the influence of multiple sequence alignment decisions that are nevertheless universally recognized as so critical to subsequent phylogenetic analysis.
Olsen concluded that: 1) we lack methods to test how heterogenous data sets are; 2) we need to confront the challenge of analyzing increasingly large data sets; 3) we have to deal with situations that appear to truly lack phylogenetic signal; and 4) the trend toward exclusively molecular analyses should be countered with efforts to integrate morphological studies more closely.
The tree of life
Starting the more taxon-based case history discussions, Piero Cammarano, Patrick Forterre and Rüdiger Cerff discussed their own ongoing efforts to use molecular data to establish relationships in the “tree of life”. Gene sequence comparisons have provided the opportunity for investigating branching orders even as ancient as more than three billion years before present, but evidence of common ancestry is often confounded with subsequent selection, lateral gene transfers, or gene duplication events. These can be exceedingly profound in their impact on phylogenetic inference, but progress in exploring highly-conserved regions of organism genomes has given those in this field optimism that deep branching patterns can be recovered.
Piero Cammarano has examined the relative strengths of different molecular probes. At issue are fundamentally conflicting hypotheses for the branching order and root position of the major bacterial lineages. Such conflicts were related to the greatly differing environments that alternative hypotheses would posit for the earliest life forms, for instance, whether the deepest ancestor of life is a hydrophilic thermophile as in Aquifex or a colder- and shallower-environment organism. Highly conserved genes that are being successfully employed include RNA polymerase subunits, EF Tu, EF G, HSPs, Tubulin, Actin, GAPDH, Glutamate dehydrogenase, tRNA synthetases, and topo-isomerases, but Cammarano especially favored multigene complexes that are coevolving, as opposed to metabolic enzymes that typically are more prone to lateral transfers and duplications. For recovering these particularly ancient branching patterns, Cammarano advocated colinear alignments of only second codon position sites that are more compositionally stable than other sites. For example, for some of these gene sequences there is typically about 18-20 % of change at second position sites contrasting with about 30 % at first position sites or 50 % at third. Second codon position sites were also found to be less subject to confounding problems of strong compositional bias.
Patrick Forterre discussed the comparative molecular biology of the ancient lineage of bacteria known as Archaea (or Archaebacteria), with some emphasis on determining the “tree of life” root from analysis of the most extensive sequence database currently available, the small subunit rRNA gene sequences, and from higher-level molecular structural features.
Forterre demonstrated that, despite frequent claims to the contrary, the rooting is far from settled. This is largely due to the inherent difficulty in distinguishing primitive versus derived features without the luxury of outgroups to life as we know it. A possible solution is to locate gene duplication events that have predated the split between these lineages. For example, several groups have rooted the tree of life in the bacterial branch using elongation factors and ATPase subunits as paralogous proteins models. However, Forterre has performed a phylogenetic analysis of elongation factors, EF-1 and EF-2, that indicates that these proteins do not contain enough phylogenetic information to root their respective trees. Such analysis is complicated by the fact that EF-2 is twice the size of EF-1, so that it is not even certain that these arose by gene duplication. In the case of ATPases, Forterre suggested that we still cannot root the tree of life because the proteins compared were probably paralogues and not orthologues. The duplication that gave rise to the two ATPase subunits was probably followed by a second duplication that led to two distinct ATPase families.
Rüdiger Cerff discussed multiple origins of gene sequences that code for aldehyde dehydrogenases (ADH) and related (e.g., GAPDH, G3PDH) proteins, which are metabolic enzymes. These are widespread in ail organisms, but Cerff’s studies of the relationships between the different classes has revealed an extremely complicated history, confounded by frequent gene duplications, lateral gene transfers, and convergent similarities due to similar selective environments. Once again, compositional biases were noted to be extreme, but these were mostly problematic at third codon positions, where G-C composition was observed to vary from about 97 % to near 0 %. One promising approach to identifying sequences as phylogenetically comparable has been to consider higher-level patterns of intron position.
Comparisons among Eukaryotic Organisms
As the focus of the participants’ study groups narrowed in taxonomy scope, the power of phylogenetic inference methods generally improved, especially when the variety and number of molecular and morphological characters increased. This was illustrated for various groups such as green algae (Russel Chapman), particular land plant families (Anne Bruneau), gastropod molluscs (Simon Tillier), echinoid echinoderms (Andrew Smith), coelomate metazoans (André Adoutte), and marsupial mammals (Arndt Von Haeseler).
Russell Chapman examined ultrastructural characters of vegetative and reproductive cells of unusual subaerial green algae (Trentepohliales), but these characters failed to resolve controversies about the class-level placement of the order. Comparison of nuclear-encoded rRNA sequences answered the question of classification, but raised new questions about the evolution of some ultrastructural characters. Similar rRNA studies of the alga Chlamydomonas, and its relatives indicate this famous genus is not monophyletic. In separate and combined analyses of morphological and molecular data, it was shown that the data sets were complementary and molecular data did not always over power less numerous morphological characters. Chapman noted that ancient, rapid radiations cannot be resolved by analysis of a stochastically changing molecule. The use of episodically changing characters such as morphological traits remains critically important in some cases.
Anne Bruneau gave examples of two particular flowering land plant (angiosperm) groups for which a variety of morphological, isozyme, chloroplast DNA, or ecological data sets were available for analysis. In a section of the plant genus Solanum, Bruneau found that partitioned data sets were incongruent with each other, and when analyzed separately yielded results whose strict consensus was poorly resolved. In contrast, analysis of the combined data set yielded a single highly-resolved tree. Bruneau argued that even given this incongruence, the combined data results should serve as the best currently available estimate of phylogeny. Another example involved the passerine bird- or hummingbird pollinated flowering plant species of the genus Erythrina. Two different data sets yielded trees that were similar in suggesting multiple derivations of hummingbird pollination, but that conflicted as to whether or not secondary shifts to passerine pollination had occurred. Trees from the combined data sets suggested a reversal to passerine within one of the hummingbird pollinated lineages was likely, supporting results obtained from a third data set.
Simon Tillier discussed gastropod phylogenetic investigations based on partial 28S (large subunit) rRNA sequences. His results contrasted with results of a recent “cladoevolutionary” analysis of gastropod morphology by Haszprunar. Experience with rRNA comparisons led Tillier to suggest that methodology is more important than nature of data. However, the rRNA results were clearly in conflict with Haszprunar’s proposed phylogeny. Tillier’s cladistic reanalysis of Haszprunar’s implied morphology-based data matrix yielded a considerably less resolved topology, but this topology was congruent with the rRNA results. One difference in the original analysis and this reanalysis was Tillier’s strict elimination of characters that included missing or inapplicable data for some of the taxa. Tillier argued that this was defensible because such characters become inconsistent statements of cladistic support.
Andrew Smith has made a variety of morphological, paleontological, and molecular comparisons among three echinoderms groups: sea urchins, ophioroids, and seastars. Smith suspects that generalities arrived at for these echinoderms are general to other organisms, but cautioned that evidence is preliminary. These empirical studies have revealed that: 1) phylogenetic problems difficult to resolve with morphological data are equally difficult to resolve with molecular evidence; 2) molecular dock assumptions are unreasonable for ribosomal genes because molecular rates by a factor of five within clades; 3) taxonomy sampling density can have a strong influence on phylogenetic analyses, both morphological and molecular, especially in the vicinity of “long branch” taxa; 4), homoplasy levels (i.e., convergences and reversals) appear slightly higher in molecular than in morphological data, but this may be due to “sanitizing” of morphological characters ; 5) usually topological differences between different data sets are a result of lack of data, not conflicting data, but sometimes different genes can provide strong but conflicting phylogenetic signals ; 6) areas of ambiguous alignment in molecular data degrade overall phylogenetic signal, and these ambiguous areas should especially be excluded as the basis for ingroup rooting decisions ; 7)-deletions in molecular sequence data are potentially informative, while treating them as unknown can generate misleading trees ; 8) rooting the ingroup is usually the least reliable part of a molecular phylogeny, but can be improved by including fossils, because one is including more proximal sister taxa.
André Adoutte and collaborators have examined the homeobox (HOM/HOX) complex in coelomate metazoans, as well as other case studies such as the “terminal” crown of eukaryotes or ciliates, and the basal lines of vertebrates or mammals. Deep polytomies or unresolved multifurcations remain a vexing problem in ail these cases. Adoutte is concerned with how to significantly establish or evaluate such branching patterns, and whether evidence for multifurcations is telling us something important about evolutionary processes. For example, rather than concluding that there is lack of resolution, we should be asking questions about the mechanisms that have driven the radiation. The manifestations of such radiations are short internal branches, unstable topologies, low bootstrap values, long branch attraction, and saturation of sites compared in molecular sequence data. Adoutte has applied Hervé Philippe’s previously mentioned JACKMONO method to go beyond simply putting bootstrap values on trees, for example to ask how many nucleotides would be necessary in order to get significant results. For the metazoan case, there is reason for optimism because there is generally congruence between molecules and morphology. However, problematic aspects remain, for example, the rRNA data support a questionable grouping of the arthropod subgroup, insects, together with nematodes, but apart from other arthropod sequences. Adoutte argued that morphological and developmental characters may be superior for such a problem because there is less chance of reversal.
Arndt Von Haeseler emphasized assumptions of molecular data analyses, with empirical examples from mammalian mitochondrial DNA (mtDNA). Von Haeseler and collaborators have recently completely sequenced an opossum (marsupial mammal) mtDNA genome, and this provides a more proximal outgroup than was previously available for establishing relationships among the six placental mammal mtDNAs published to date. Von Haeseler pointed out that several commonly employed methods make strong assumptions that are clearly violated in reality. For example, Kimura’s two-parameter model, in which transition and transversion substitutions are considered separately, assumes that nucleotides occur with equal frequency (they do not, and they differ by codon position), that the model employed is the same for all sequences (this is highly unlikely), and that ail sites are independent (violations of independence are known). Von Haeseler discussed approaches to improving the model and method of inference employed, and compared how well different genes performed in comparison to predicting the results obtained from most other genes. A major conclusion was that first, second, and third codon positions have to be analyzed separately. While the evolutionary rates of first and second codon positions are low, the comparison of pairwise differences of third codon positions indicates that they are already saturated. Ironically, these tendencies are especially pronounced in the gene that codes for cytochrome b proteins, the one that has been compared most intensely, making it among the poorest choices of mtDNA genes available. Von Haesler still cautioned against excluding third codon positions from consideration as long as at least some are not saturated and assuming there is no hierarchical pattern in the substitutions subsequent to lineage splitting. As was demonstrated by overall congruency when only third position sites were analyzed, significant-appearing saturation does not completely swamp out existing phylogenetic structure.
Insects as Model Systems
Several participants discussed examples of employing specific insect groups to better understand molecular evolutionary processes. Fred Breeuwer began with a discussion of the important role that symbiotic (i.e., commensal or parasitic) microorganisms can play in the evolution of host organisms, with specific examples of bacterial symbionts with insect (or other arthropod) hosts. These symbionts are widespread and can affect the host insect’s meiotic or mitotic reproductive processes. These have tremendous relevance to studies of insect molecular evolution because the symbionts can promote lateral gene transfers, which can confound attempts to reconstruct phylogenies. In some cases, they could be responsible for explaining rapid radiations in particular host lineages. Moreover, the symbionts can also directly influence the way that natural selection acts. Breeuwer gave many examples that suggested that an insect might really be a slave to the symbiotic microorganisms. These symbionts can either be mutualists (e.g., endosymbionts, sulfur-oxidizing microbes, nitrogen fixing bacteria) or selfish (e.g., sex changers, male killers, cytoplasmic incompatibilities). The selfish symbionts can effectively control reproduction and sex determination in their hosts. There has been recent progress in applying PCR based sequencing methods in order to reconstruct phylogenetic relationships of the various symbionts. This has provided evidence that symbionts of particular functional types tend to be related by descent. Parallel studies of hosts and symbionts have indicated a high level of congruence between host and mutualistic symbiont phylogenies, but not between host and selfish symbiont phylogenies. The latter might indicate a high level of lateral gene transfer.
Costas Krimbas summarized the considerable progress in genetic understanding of drosophilid flies, whose giant chromosomes have been the subject of exhaustive studies. Krimbas emphasized studies that have attempted to reconstruct a succession of chromosomal events, employing band staining techniques as a landmark-based pattern recognition System to distinguish between different regions with considerable accuracy. Krimbas argued that this ability to track characters is not available in many molecular sequence character comparisons, and thus correspondence is not always as easily demonstrated for sequence data. Examples given date back to classic studies by Dobzhansky and others, including Krimbas. One involves Drosophila pseudoobscura, D. miranda, and D. persimilis, which differ in gene orders. Krimbas observed that, while it is difficult to know which gene order is first, it is most “economical” to place D. pseudoobscura between the other two, because a transition to either of the other two orders can be explained by a single step. Krimbas updated Dobzhansky’s view that chromosomal inversion events are so rare that they can be treated as unique. Given modem knowledge of chromosomal hot spots and transposable elements, such inversions are now considered rare but not unique. However, absence of “creeping” or sustained polymorphism beyond species formation or absence of evidence of introgression can be used to infer that fixation in phylogeny occurred only once.
Rob DeSalle extended these points with his own examples of studies on Drosophila, which integrate cladistic analyses, genetic studies, and molecular approaches. The Drosophila model system is ideal because morphological, molecular, developmental, and cellular understanding is in an advanced stage. DeSalle intentionally overstated a view of flies as bags for development of bristles, heads, wings, segments, and genitalia, because this leads to specific tests of developmental hypotheses. For example, variation in bristle pointing direction, with bent over bristles known as the “bb” or “bobbed” trait, has been mapped to the ribosomal gene complex and linked to a reduction in number of cistrons in the complex. Using a combined evidence approach to phylogeny reconstruction, DeSalle was able to demonstrate that bb has most likely arisen multiple times, i.e., cistron reduction is common. Congruent with this conclusion, cistron reduction in two species was shown to have occurred in different gene areas. DeSalle and collaborator have shown that the reduction in cistron inserts is necessary, but not sufficient. A third species was shown to involve another gene locus that appears to regulate bb. As illustrated with this and several other examples, DeSalle’s approach starts with a cladogram and associated character hypotheses or tests of homology, thon leads to the identification of anomalous characters or homology assessments, then assesses applicability of “phyletic phenocopy” (or apomorphic tendency), and finally leads to character dissection. This involves, in the best case, obtaining primary sequences, observing the distribution of gene products in embryos and imaginal discs, and demonstrating interactions that occur “downstream” from tire initial expression of control gene products.
European Systematics in the Twenty-first Century
The final session of the workshop addressed the goals of ESF as they relate to fostering improvements in the European systematics community. The many historical strengths of European systematics are challenged by increasingly sparse funding opportunities and political or cultural barriers to communication.
Simon Tillier summarized the goals of the ESF. It is recognized that systematics is a fundamental necessity; without it there is no biology. There is also a circumstantial necessity, given the trend towards declining systematic knowledge, due historical importance of European systematic collections, and its pivotal role in the organization and subsidizing of science in Europe. There is also great opportunity, given the present interest in preserving biodiversity and the rapid growth of molecular and theoretical systematics as disciplines.
Workshop participants engaged in a lively discussion about how the interrelationships between systematics, taxonomy, and evolutionary process studies and funding opportunities are perceived, especially by the public and political funding agencies. It was recognized that there is need for a taxon based emphasis in systematics research, but it is critical that taxon-based researchers demonstrate to the biological community why we need systematics research, and its basic relationship to biological research. The reality of many academic institutions is that research specialties viewed as stagnant or old fashioned are negatively impacted. There was a strong consensus that these workshop presentations demonstrated that systematics research is very current and exciting. As such, the workshop was a successful vehicle for ensuring the continued health and prominence of European systematics.