Or Tete




right r;/ Tntmdftlint ft mcrrixL

1.22. The title page of Darwin's On the Origin of Species (1859). The first edition of Darwin's book contains a single figure - a diagram of a phylogenetic tree. Photo: AMNH Library.

to be changed because the findings of systematists created evolutionary theory in the first place. The theory of evolutionary change by means of natural selection merely provided an explanation, a theoretical scaffolding, for the variation that systematists studied. Darwin himself concluded that all natural classifications produced by systematists are genealogical, and, if reconstructed carefully, they are evolutionary classifications. In fact, though, the theory of natural selection explained anagenetic change, or the evolution of particular characters, not the origin or formation of new species. This would become a major issue in the twentieth century, in which entomologists had a substantial impact.

After Darwin

Many biologists adopted evolutionism after the publication of The Origin of Species, but some influential biologists resisted

1.21. Alfred R. Wallace (1823-1913), adventurer, naturalist, and prolific insect collector. Wallace was coauthor with Darwin on the original paper proposing evolution by means of natural selection. Photo: AMNH Library.

the concept of natural selection, like the great British anatomist Richard Owen (see Rupke, 1994), and some even opposed the very thought of evolution (Agassiz, 1896). From 1905 to 1920 Thomas H. Morgan's "fly lab" at Columbia University discovered scores of important genetic phenomena in Drosophila, which won Morgan the Nobel Prize, but even he denounced the significance of natural selection. Population and quantitative geneticists in the 1920s - R. Fisher, J. B. S. Haldane, and S. Wright - took the new genetics and applied it to explanations of evolutionary change, thus starting the "New Synthesis," the marriage of genetics and systematics. To them, mutation was the source of raw material for evolutionary change, the clay, and natural selection the creative force, the sculptor. Theodosius Dobzhansky, who was originally a beetle systematist, was impressed by the genetic and systematic work of Morgan's insightful student, Alfred Sturtevant, and applied studies of wild Drosophila to questions of genetic variation in nature and the formation of species (Dobzhansky, 1937). A bird systematist at the American Museum of Natural History, Ernst Mayr, mined systematic studies in the tradition of Darwin and proposed the very influential idea of allopatry: Species evolve from geographical isolation (Mayr, 1942). Another American Museum scientist, the paleontologist George G. Simpson, added the geological perspective to the New Synthesis by concluding that the gradual change observed in the fossil record was explained by small, incremental genetic changes (a concept challenged 30 years later by Eldredge and Gould's punctuated equilibrium). Between the 1940s and 1960s, biology was preoccupied with the mechanisms of evolutionary change, and while the New Synthesis relied on the work of systematists, systematics per se was marginalized. Ironically, during the foment of the New Synthesis, the brilliant geneticist Alfred Sturtevant was quietly working away on the systematics of drosophilid flies. The methods he used in his classification of the genus Drosophila (Sturtevant, 1942) actually presaged the theoretical and empirical work of another highly influential insect systema-tist, Willi Hennig. Sturtevant's paper was overlooked, perhaps because it was published the same year as Mayr's highly influential Systematics and the Origin of Species.

In the 1950s and 1960s scholars began to address these deficiences, particularly attempting to get at an objective means for reconstructing relationships among taxa. One response was the development of numerical taxonomy, or phenetics. Developed principally at the University of Kansas by the entomologists Robert Sokal (Figure 1.23) and Charles D. Michener and by the microbiologist Peter Sneath from Britain, phenetics attempted to develop a method of grouping organisms based on their overall similarity (e.g., Michener and Sokal, 1957; Sokal and Sneath, 1963; Sneath and Sokal, 1973). The goal was to make classifications logical, repeatable, and methodological, in an attempt to remove as much subjectivity as possible. It also required users to record

1.23. Three prominent entomologists and major architects of three philosophies of systematic thinking - from left to right, Willi Hennig, fly systematist and founder of phylogenetics; Robert Usinger, bug systematist and proponent of evolutionary taxonomy; and Robert Sokal, fly population geneticist and cofounder of phenetics. Photo: G. W. Byers, University of Kansas Natural History Museum.

1.23. Three prominent entomologists and major architects of three philosophies of systematic thinking - from left to right, Willi Hennig, fly systematist and founder of phylogenetics; Robert Usinger, bug systematist and proponent of evolutionary taxonomy; and Robert Sokal, fly population geneticist and cofounder of phenetics. Photo: G. W. Byers, University of Kansas Natural History Museum.

as many differences as could be measured among species, which was another advantage over previous methods of classification. Rather than presuming that one or a few characters have particular evolutionary importance, phenetics attempted to obtain a holistic picture of diverse attributes of taxa. One recited advantage of the approach was that it offered an explicit and testable methodology, but explicitness and testability per se (criteria later espoused by cladists, below) are the most minimal of scientific criteria. An absurd hypothesis ("the moon is made of green cheese") can be very testable. Phenetics had other problems too. First, the choice of specimens and characters to measure imposed a substantial bias, particularly since different systematists considered different characters to be important for study, and a change in a few characters could yield dramatically different results. Also, there was an overall loss of information, such that characters were summarized in a similarity matrix, so the significance of individual characters was hidden in the statistical wash (but this problem is again resurfacing in some molecular analyses). Similarly, different analytical methods gave radically different results, so the classifications turned out not to be stable. Additionally, the predictive value of the phenetic classifications was low, and thus obscured evolutionary patterns, which is because phenetics measures dissimilarity, not relationships. In a phenetic scheme, an insect species that lost its wings might be classified in a genus or family completely different from a closely related winged species. The overarching problem with the use of phenetics is that biologists are interested in phylogeny and require classifications that reflect natural, genealogical relationships. Rather than estimate phylogenetic relationships, the method measured phenetic similarity, failing to distinguish between evolutionary similarity (the result of descent) from convergence and that from parallelism. Unfortunately, even today, some phe-netic analyses (masked under the name of distance methods) are erroneously used to obtain what are believed to be genealogical relationships. Until DNA sequencing became available, bacteriologists relied heavily on phenetic methods because their organisms lack numerous useful morphological features. Other biologists, however, largely abandoned phenetic methods or never accepted them because such classifications consciously ignored phylogeny. Various authors tinkered with the method in an attempt to make it more representative of phylogeny, but the results were not persuasive. As a result of these problems, pheneticists adopted a nominalist philosophy in which only individuals exist and all other groups are artifacts of the human mind. Thus, the classification of plants and animals could be treated the same as inanimate objects in this system. Now, species are classified as lineages, groups that are closely related by common descent.

Roughly contemporaneous with the development of phe-netics, another response to the NeoDarwinists was being pursued, but this one was more fruitful. Willi Hennig (1913-76) (Figures 1.23, 4.21), a German fly systematist, operated under the principle that phylogeny can be reconstructed and that all classifications should be based on the reconstructed pattern of genealogical descent, thus echoing Darwin's original call for evolutionary, or natural, classifications. Hennig provided a rigorous method for analysis of phy-logenetic relationships, called cladistics (Hennig, 1965b, 1966). The basic idea was that a truly phylogenetic system of classification would be the most useful as a general reference system for biology. If a classification is to reflect phylogeny exactly and have explanatory power, then all taxa classified must be strictly monophyletic. Paraphyletic and polyphyletic groups are not accepted as taxa suitable for ranking in the classification (see also Farris, 1974, for characterization of these terms). Monophyletic groups were natural because they contained the common ancestor and all the descendent species. Paraphyletic groups contained the common ancestor and some, but not all, of its descendent species. The most problematic groupings are polyphyletic, which are those that contain some of the descendants of a common ancestor but not the common ancestor itself (Figure 1.24).

These distinctions serve to highlight another of Hennig's major contributions, the concept that among the mosaic of an organism's characters only those that are derived ("specialized," "modified") and shared with other species are informative of phylogeny. Hennig noted that, for the purposes of phylogenetic reconstruction, there are actually two kinds of character similarity. Apomorphies are similarities that arose in a most recent common ancestor, or a recently evolved ("advanced") feature that appears only in a group of closely related species. Plesiomorphies are similarities that

1.24. Three kinds of groupings - monophyletic, paraphyletic, and polyphyletic. Natural classifications attempt to categorize and name only monophyletic groups.

arose in a distant common ancestor, or "primitive" feature. Use of "advanced" and "primitive" is generally discouraged because these terms imply a scale of perfection or adaptation, while plesiomorphic features are merely apomorphic features deeper in phylogeny. Wings are a plesiomorphy of butterflies, but an apomorphy of pterygote insects. In other words, it is important to note that "plesiomorphy" and "apo-morphy" are relative terms.

Thus, the critical aspect of recognizing monophyletic groups was to base them on shared apomorphies (called synapomorphies) and not on shared plesiomorphies (called symplesiomorphies). Grouping taxa based on plesiomorphies would only produce paraphyletic groups. A group is mono-phyletic if it is characterized by the possession of synapomor-phies, derived characters shared by the members of the group. For example, wings are a synapomorphic trait that unites the winged insects into a monophyletic group called Pterygota. Taxa that are paraphyletic are based on plesiomor-phies. In the same example, before the evolutionary novelty of wings arose, insects were primitively wingless. Thus, the absence of wings is plesiomorphic for insects and the aptery-gotes (wingless insects) are paraphyletic, as some apterygote taxa are actually more closely related to Pterygota than they are to other wingless species. Lastly, polyphyletic groups are frequently based upon convergent characters that are only superficially similar but not of a common evolutionary derivation. The sucking insects, for example, are a polyphyletic assemblage, the development of such mouthparts having arisen several times and in different configurations. Hennig thereby provided us with the framework to define characters and then to polarize them into apomorphic and plesiomor-phic states. As already discussed, the critical aspect of recognizing monophyletic groups was to base them on shared apomorphies and not on shared plesiomorphies. The presence of an apomorphy in a single species and in no other species is not informative of relationships because it groups the taxon with no other lineage. These are called autapomorphies.

However, how does one know whether a character is apomorphic or plesiomorphic? This is the problem of determining character polarity, and various methods have been proposed to determine this, the most widely used being out-group comparison. Lineages closely related to the one under study but sharing traits through a more ancient common ancestor (the outgroup) share plesiomorphies; those traits that are unique within the ingroup are apomorphies. Other, less often employed or defunct methods involve criteria based on ontogeny or development, frequency of characters, and paleontology. The ontogenetic criterion uses developmental evidence, which sometimes works but is rarely practical. The frequency criterion states that the most common character state is plesiomorphic. This can be very misleading because groups that recently radiated will have the most species with derived features and only a few species will remain with plesiomorphic features. The paleontological criterion maintains that the characters found in the oldest fossils are plesiomorphic. This can also be misleading because fossil taxa often have some very highly modified traits, as we show throughout this book.

Cladistic analyses are all about congruence. Each character is allowed to influence the final set of relationships, unlike in phenetics where the effects of individual characters are obscured and irretrievable. We then need to find the overall hierarchical pattern implied by the various similarities. Before we elaborate on finer aspects of the methodology, however, we must step back for a moment and consider what these similarities really are.


Early in the history of comparative biology, scientists recognized a general similarity among some organisms. Common features could be recognized in an overall body plan, but scientists could not explain why. Organisms seemed to be made of similar parts organized into a similar body plan with simple modifications at various points that created diversity, a phenomenon that required explanation. One of the most famous examples of the early recognition of this similarity in body plans is by Belon (1555), who published a book on comparative natural history in which he noted various topologi-cal similarities in the skeletons of a human and a bird. He identified, based on relative positions and connections, "identical" bones in the common tetrapod body plan.

Various naturalists, particularly vertebrate anatomists, played with this idea, and it formed the basis for all comparative biology. The next major development in the concept of homology was not until the early 1800s, when Etienne Geoffroy Saint-Hilaire (1818) at the Paris Museum wrote his PhilosophieAnatomique (Figure 1.25). Geoffroy used what he called the principle of connections. Based on the relative connections of structures in the overall body plan, he identified similar structures even if they had been slightly or dramatically modified between two organisms. He went so far as to

1.25. The great French anatomist and natural philosopher, Etienne Geoffroy Saint-Hilaire (1772-1844). Geoffroy laid the foundation for recognizing homologous traits, although he personally employed the term "analogy" for what we today call homology. Photo: © Bibliothèque Centrale MNHN Paris 2003.

1.25. The great French anatomist and natural philosopher, Etienne Geoffroy Saint-Hilaire (1772-1844). Geoffroy laid the foundation for recognizing homologous traits, although he personally employed the term "analogy" for what we today call homology. Photo: © Bibliothèque Centrale MNHN Paris 2003.

apply his method throughout the vertebrates and developed in his book the idea that the inner ear bones were in fact bones modified from part of the jaw in lower vertebrates, a revolutionary thought at the time. Many people ridiculed him (particularly his prestigious adversary Cuvier), but an origin of the vertebrate ear bones from jaw bones is entirely accepted today. This principle of connections allowed for the recognition of homologies, common structures in a common organismal plan (although Geoffroy himself used the term "analogy" for what we today call "homology").

Richard Owen (1866) expanded on Geoffroy's concept and noted that some structures might appear superficially similar but that, based on the principle of connections, they were indeed not identical in the common body plan. He therefore called these "deceptive" characters analogies. He, being a bit of a fan of Plato, more fully developed the idea of an archetype, that there was some "ideal" form and that all structures in organisms were modifications of God's ideal. Comparing any organism to the archetypal ideal would allow for a meaningful comparison between the degenerate forms observed on Earth. Owen, like his predecessors, however, still explained homologies as evidence of the hand of God, which is one reason why he and Darwin clashed.

Darwin (1859) once again provided the critical synthesis recognizing this observational phenomenon of topological similarities (i.e., homologies) among organisms and how these formed the basis for defining hierarchical groups. Thus, if the observed hierarchy was explained as a product of evolutionary descent, then the observed homologies shared among organisms must be shared because of their ancestry. In other words, organisms share traits because their common ancestor had them. Darwin's contribution was simply a matter of interpretation and explanation, but it was critical and fundamental.

Darwin was quick to point out a logical separation for the observation of homology versus its explanation via ancestry, just as for the hierarchy of life and its explanation by descent. His vehement follower Huxley, however, unified homology and ancestry such that, in a single term, we had the observation and its explanation. Today, we tend to forget the observational basis of homology and focus simply on the fact that it is a trait shared between two organisms because they inherited it through their most recent common ancestor. The observational phenomenon and the explanation of that observation are often combined and confused, but there is indeed such a distinction within the single concept of homology, which is critical for keeping evolutionary theory from becoming logically circular. Although we interpret homology as a shared trait inherited from a common ancestor, we do not observe common ancestors. We need criteria for recognizing homol-ogy or rules that allow us to hypothesize why two or more species are closest relatives (Brady, 1985, 1994).

The modern definition of homology is basically an equation with synapomorphy. This unification is fine as long as one keeps in mind the logical separation between the observation of homology and the demonstration of ancestry, the practice and the conclusion. To distinguish observational homology from that equal to synapomorphy, some authors have employed the term "primary homology" for the former and "secondary homology" for the latter (e.g., de Pinna, 1991). Primary homologies are observational homologies not yet tested by a cladistic analysis, while secondary homologies are synapomorphies (i.e., those tested by a cladistic analysis and interpreted to be the product of ancestry).

Many people have gone about characterizing criteria for observational homology (i.e., primary homologies). The most extensive work was that of Remane (1952), who provided three "criteria of similarity" that can be used to identify homologous features.

1. Criterion of Position. A structure in different species that has a similar position relative to other landmarks in the body plan is likely to be homologous (this is essentially the topological identity criterion of Geoffroy Saint-Hilaire). This is also the main criterion (and perhaps the most important) for molecular homology because identifying a base-pair substitution depends on what base-pairs flank it. In morphology, too, this is a fundamental criterion. For example, the criterion of position allows one to readily identify that stalked eyes are not homologous across various families of flies. An important landmark for these topological comparisons is the position of the antennae (Figure 1.26). In the Diopsidae, for example, the antennae are positioned at the apex of the stalks and near the compound eyes; in other families, such as various tephritoid flies, they are normally positioned near the midline of the face. The position of the antennae is the clue that the "stalk" is composed of paraocular integument in tephri-toids but also of the frons and face itself in Diopsidae (Figure 1.26). The fossorial legs of many insects is another excellent example where topological identity allows for the discernment of homologous versus nonhomologous traits (Figure 1.27).

2. Criterion of "Special Similarity." Complex structures that agree in their details are more likely to be homologous. An excellent example concerns the reduced, club-like wings in flies (Diptera) and the small order Strepsiptera. Some molecular evidence suggests that these orders are closely related (Whiting et al., 1997; Wheeler et al., 2001), but the halteres of flies are the hind wings (Figure 12.23), and those of Strepsiptera are forewings (Figure 10.79). Whiting and Wheeler (1994) were so convinced by the "special similarity," essentially shape, of the halteres, that they proposed a developmental switch of the mid- and hind thoracic segments in the two orders (see Chapter 10). For them, the criterion of "special similarity" overrode the criterion of position.

3. Criterion of Continuation. Structures in two or more taxa may be quite dissimilar, but if other taxa can be found that have intermediate forms so that dissimilar extremes can be linked, then the homology may be more readily recognized. For example, the proboscis of glossatan moths is usually very long and coiled and is formed from extremely long galeae (Figure 13.19). It is possible that the galeae could not be recognized if early intermediates between glossatan and mandibulate moths were not known.

Modifications of these criteria have been adapted for the identification of behavioral and molecular homologies as well (e.g., Baerands, 1958; Wenzel, 1992; Greene, 1994; Hillis, 1994; Miller and Wenzel, 1995). Interestingly, these two types of characters are somewhat related by the means in which homologies are identified.

Molecular data principally consist of identifying the sequence of amino acids (for proteins) or nucleotides (for DNA) from a particular enzyme or gene. This produces a linear sequence of point-by-point identifications. In order to compare two or more sequences, they must first be aligned so that corresponding points in the sequences are paired. Although this sounds trivial on the surface, it is, in fact, often

1.26. An example of homologizing structures using the criteria of size, position, and "special similarity": the heads of stalk-eyed flies in different families. Diopsid antennae are near the end of the stalks; this and the boundaries (sutures) between the plate-like sclerites indicate that the face of diopsids is expanded. In the other flies, the frontoorbital plates on the inside margin of each eye are expanded, and antennae remain close together on the face. Not to the same scale.

Mronto-genal fissure

Otitidae (Plagiocephalus) antenna

1.26. An example of homologizing structures using the criteria of size, position, and "special similarity": the heads of stalk-eyed flies in different families. Diopsid antennae are near the end of the stalks; this and the boundaries (sutures) between the plate-like sclerites indicate that the face of diopsids is expanded. In the other flies, the frontoorbital plates on the inside margin of each eye are expanded, and antennae remain close together on the face. Not to the same scale.

difficult and nonintuitive. Deletions or insertions into the sequence of one or more taxa create linear sequences of unequal lengths. Thus, to make the sequences align, gaps indicating where the deletion or insertion took place may have to be hypothesized so that the remaining amino acids or nucleotides are aligned. Gaps, however, are not observed; they are hypothesized ad hoc because a strand of DNA does not possess a physical gap, just a lost section of nucleotides. For regions of DNA that code for proteins, the codon structure provides a nice frame of reference for aligning sequences. However, noncoding regions can be problematic, with numerous insertions and deletions. Once again, this is a critical first step for molecular analysis because it is essentially the identification of primary homologies. Each nucleotide position is a character in the analysis; consequently, alternative alignments, which compare nonidentical positions across the same strands, may arrive at wildly different sets of relationships, skewing our interpretation of evolutionary history. Whereas in morphology the last two of Remane's three criteria can be called upon to refine our hypothesis of observed homology, molecular data rely almost solely on topological position.

The recognition of behavioral homologies can be analogous to the identification of molecular homology because it is often possible for a behavioral repertoire to be taken apart into a sequence of acts. For instance, the mating displays of some flies are ritualized performances of various acts; the male may have to wave his wings, move around the female, and tap her, in a specific manner and order before she will allow mating. Thus, like molecular data, a linear sequence of acts must be aligned when comparing two or more species. Furthermore, like molecular data, hypothesized gaps may be required. For example, in the mating ritual, one species of a genus may have lost the wing-flipping motion seen in all the other species and, instead, skip directly to the final act of tapping. In this example, a gap would need to be inserted to align properly the behaviors observed. In contrast to molecular sequences, however, the last two criteria for homology can also be applied to behavioral repertoires. Behaviors need not always appear as a repertoire and can extend beyond the organism itself. Several insects have diagnostic behaviors that result in the alteration of the environment (e.g., feeding damage) or the production of some lasting structure (e.g., nests). In some instances, different lineages produce diagnostic types of environmental changes that can frequently be preserved through geological time and provide us with early records of behavioral novelties. Leaf-cutter bees produce a diagnostic form of leaf damage often seen in Tertiary fossil leaves (e.g., Wappler and Engel, 2003). Nests are also commonly preserved, leaving behind evidence of their ancient architects. Fossilized behaviors provide important insights into the stages of evolution. Even though many people a priori believe that behavior is too labile to be informative for phylogeny, it is in fact no different from other types of character data. It is best never to assume a priori that a feature will not be phylogenetically informative because only a cladistic analysis can determine this. Indeed, some behaviors are highly variable and do not explain relationships, but in the same way that some morphological traits or gene regions can also be too labile or variable. In studies comparing the use of behavioral characters in cladistic analyses, behaviors were found to be just as informative as other kinds of data (e.g., Wenzel, 1992; de Queiroz and Wimberger, 1993; Wimberger and de Queiroz, 1996).

1.27. A classic example of convergence among insects: fossorial forelegs. Forelegs adapted for digging have evolved repeatedly in insects, but in each case the forelegs have been modified in different ways. Not to the same scale.

Unlike morphology or behavior, with their seemingly endless variety of comparable forms, the criteria for analyzing molecular data are more limited. Though extensive stretches of DNA sequences are easily gathered, each character is confined to only four possible states: A, T, C, and G. The language of molecular systematics is short of words. This limits the kinds of change that may take place. If multiple changes have occurred at the same position, the probability of arriving at the same nucleotide by chance alone and not by descent increases. Such multiple "hits" at a nucleotide position essentially erase the historical information that was once stored in the DNA sequence and can be difficult to discern. Because DNA sequences rely solely on Remane's first criterion of homology, multiple "hits" often lead to erroneous homologies. With no recourse to the criterion of "special similarity" (after all, an adenosine is an adenosine), nonhomologous nucleotides in identical positions can mislead in an analysis by uniting two or more species that, in actuality, are not closely related at all. The more multiple hits across the entire sequence, the more random the final inference of phy-logeny becomes. This problem is avoided to some extent by sequencing genes that have appropriate substitution rates. Genes with high substitution rates ("fast evolving" genes) are used for species-level relationships; genes with low rates ("slow evolving") are used for more ancient divergences. Specialized analyses have also been designed in attempts to confront and solve this difficulty.

In reconstructing relationships, molecular data presently draw from a more limited portion of the genome than the information derived from morphology relies upon. Most morphological structures in insects are highly polygenic, so even the most cursory morphological data set may represent the cumulative effect of hundreds to even thousands of genes from across the genome. For example, the shapes of several male structures of drosophilid fruitflies, each of which comprises one or two systematic characters, are controlled by between four and ten genes (Templeton, 1977; Val, 1977; Coyne, 1983). Presently, the largest molecular studies may provide data from about eight genes, usually fewer, and reflect a much more restricted part of the genome. This is particularly true given that most studies, regardless of taxon, focus on a relatively small suite of genes for phylogenetic inference (Table 1.2). Moreover, the phenotype not only reflects genotype, but it is also an emergent phenomenon, comprised of the interaction of genotype and environment. Naturally, as sequencing technology and computational methods improve, this setback should be overcome, but there will always be a need for studying morphological characters, particularly when interpreting fossils.

Phylogenetic Analyses

At its simplest, reconstructing phylogeny boils down to a congruence test among topological identities, that is, of either morphological, behavioral, or molecular homologies. The observed homologies are analyzed cladistically - divided into apomorphies and plesiomorphies to form a hierarchical pattern, a cladogram. The cladogram is a type of very general evolutionary tree that indicates only relative relationships, not ancestor-descendant relationships. A cladogram calibrated with the fossil record and the geological time scale is considered a phylogeny (Smith, 1994). After this has been completed, the pattern of change of the individual characters can be interpreted. Homologies that support monophyletic groups are interpreted as synapomorphies. Alternatively, homologies that appear independently in different places on the phylogeny are interpreted as homoplastic (= analogies). In other words, the trait is incongruent with the hierarchical

+1 0

Post a comment