A Reconsideration of the Coding of Inapplicable Characters: Assumptions and Problems

Character coding entails assumptions that may be problematic within the context of parsimony analysis using current computer algorithms. The example discussed here involves a character-variable (e.g., tail color) that is inapplicable in some taxa in the analysis because the part (e.g., tail) with which it is associated is lacking in those taxa. The part and character-variable can be coded as separate characters, or they can be fused into a single character. If the part and character-variable are coded as separate characters there is transformational independence between the part and the character-variable, but the logical dependence inherent to the hierarchical relationship between the part and its character-variable is only partly accounted for. Fusing the part and character-variable into one multistate character fully accounts for the logical dependence, but it is equivocal regarding the transformational independence. Separate coding is consistent with the primary homology statement that the part is homologous in all taxa possessing it, whereas fused coding is equivocal regarding this hypothesis of primary homology. As a result fused coding involves a loss of phylogenetic information. Use of a stepmatrix or other mechanisms associated with fused coding that preserve this phylogenetic information involves weighting schemes or ordered characters that have other assumptions that may also be difficult to justify.     

Artifacts of Coding Amino Acids and Other Composite Characters for Phylogenetic Analysis

A phylogenetic analysis can be no better than the characters on which it is based. Just as it is inappropriate to code character states of individual characters as separate presence/absence characters, it is inappropriate to combine independent characters because not all information in the data is being utilized. Composite characters link otherwise discernible states from different characters together to form new character states. There are two related problems with this coding. First, there is a loss of hierarchic information between the reductive and composite characters when unordered states are used. Second, the linking of separate characters that occurs during the construction of composite character states can create putative synapomorphies that were not present in the separate characters. For amino acid characters, the problem may occur whenever more than one position of a codon is variable among the terminals sampled. Groups that are resolved as paraphyletic with reductive coding may be resolved as monophyletic with composite coding. The artificial character states indicated by the amino acid characters are unlikely to be congruent with the true gene tree.     

EVOLUTIONARY CHARACTER ANALYSIS: TRACING CHARACTER CHANGE ON A CLADOGRAM

Abstract— There has been little formal discussion concerning character analysis in cladistics, even though characters and their character state trees are central to phylogenetic analyses. We refer to this field as Evolutionary Character Analysis. This paper defines the components of evolutionary character analysis: character state trees, transmodal characters, cladogram characters, attribute and character phylogenies; and the use of these components in phylogenetic inference and evolutionary studies. Character state trees and their effect on cladogram construction are discussed. A new method for numerically coding complex character state trees is described that further reduces the number of variables required to describe them. This method, ordinal coding, reduces the size of data matrices, and facilitates retrieval of state codes. This paper advocates the use of both biological evidence and evidence internal to the cladogram itself to construct character state trees (CSTs). We discuss general models of character evolution (morphocline analysis, Fitch minimum mutation model, etc.) and their role in forming CSTs. Character state trees formed with theories of character evolution are referred to as transmodal characters. These transmodal characters are contrasted with cladogram characters (Mickevich, 1982), and the place of each in a phylogenetic analysis is discussed. The method for determining cladogram characters is detailed with more complicated examples than found in previous publications. We advocate testing transmodal characters by comparing them with the resultant cladogram characters. This comparison involves transformation series analysis (TSA; Mickevich, 1982) which is viewed as an extension of reciprocal illumination. The TSA procedure and its place in hypothesis testing are reviewed. Tracing the evolution of characters interests both systematists and non-systematists alike. When character state trees (transmodal characters) are optimized on pre-existing phylogenies, character phylogenies and attribute phylogenies result. Attributes are defined as a feature that may or may not be homologous (i.e., ecological categories, plant hosts, etc.). We provide two illustrations of this approach, one involving the evolution of the anuran ear and another involving the coevolution of the butterfly Heliconius and its hostplants. Finally, the components of phylogenetic character analysis can be used to test more general evolutionary theories such as the biogenetic law and vicariance biogeography.     

When molecules and morphology clash: reconciling conflicting phylogenies of the Metazoa by considering secondary character loss

Molecular and morphological data sets have yielded conflicting phylogenies for the Metazoa. So far, no general explanation for the existence of this conflict has been suggested. However, I believe that a neglected aspect of metazoan cladistics has introduced a systematic and substantial bias into morphological phylogenetic analyses. Most characters used for metazoan cladistics are coded as binary absence/presence characters. For most of these characters, the absence states are assumed to be uninformative default plesiomorphies, if they are defined at all. This character coding strategy could seriously underestimate the number of informative apomorphic absences or secondary character losses. Because nodes in morphological metazoan phylogenies are typically supported by relatively small numbers of characters each with a potentially strong impact on tree topology, failure to distinguish between primary absence and secondary loss of characters before a cladistic analysis may mislead morphological cladistics. This may falsely suggest conflict with molecular phylogenies, which are not sensitive to this bias. To test the existence of this bias, I compare the phylogenetic placement of a variety of metazoan taxa in molecular and morphological trees. In all instances investigated here, phylogenetic conflict can be resolved by allowing for secondary loss of morphological characters, which were assumed to be primitively absent in cladistic analyses. These findings suggest that we should be cautious in interpreting the results of morphological metazoan cladistic analyses and additionally illustrate the value of a more functional approach to comparative morphology in certain circumstances.     

The Logical Basis for the use of Continuous Characters in Phylogenetic Systematics

It has been argued that continuous characteristics should be excluded from cladistic analysis for two reasons: because the data are considered inappropriate; and because the methods for the conversion of these data into codes are considered arbitrary. Metric data, however, fulfill the sole criterion for inclusion in phylogenetic analysis, the presence of homologous character states, and thus cannot be excluded as a class of data. The second line of reasoning, that coding methods are arbitrary, applies to gap and segment coding, but quantitative data can be coded in a nonarbitrary manner by means of tests of statistical significance. These procedures, which are both objective and repeatable, determine the probability that two taxa possess an homologous character state; that is, if they have inherited a particular central tendency and distribution of individual variates unchanged from a common ancestor. Thus, the application of statistical tests to quantitative data empirically detects the presence of evolu tionary change, the raw material of phylogenetic reconstruction.     

The systematic utility of floral and vegetative fragrance in two genera of nyctaginaceae

We examined relationships between fragrance and phylogeny using a number of approaches to coding fragrance data and comparing the hierarchical information in fragrance data with the phylogenetic signal in a DNA sequence data set. We first used distance analyses to determine which coding method(s) best distinguishes species while grouping conspecifics. Results suggest that interspecific differences in fragrance composition were maximized by coding as presence/absence of fragrance compounds and biosynthetic pathways rather than when quantitative information was also included. Useful systematic information came from both compounds and pathways and from fragrance emitted by both floral and vegetative tissues. The coding methods that emerged from the distance analyses as best distinguishing species were then adapted for use in phylogenetic analysis. Although hierarchical signal among fragrance data sets was congruent, this signal was highly incongruent with the phylogenetic signal in the DNA sequence data. Notably, topologies inferred from fragrance data sets were congruent with the DNA topology only in the most distal portions (e.g., sister group pairs or closely related species that had similar fragrance profiles were often recovered by analyses of fragrance). Examination of consistency and retention indices for individual fragrance compounds and pathways as optimized onto one of the most-parsimonious trees inferred from DNA data revealed that although most compounds were homoplastic, some compounds were perfectly congruent with the DNA phylogeny. In particular, compounds and pathways found in a few taxa were less homoplastic than those found in many taxa. Pathways that synthesize few volatiles also seem to have lower homoplasy than those that produce many. Although fragrance data as a whole may not be useful in phylogeny reconstruction, these data can provide additional support for clades reconstructed with other types of characters. Factors other than phylogeny, including pollinator interactions, also likely influence fragrance composition.     

Measuring the phylogenetic randomness of biological data sets

Two qualitative taxonomic characters are potentially compatible if the states of each can be ordered into a character state tree in such a way that the two resulting character state trees are compatible. The number of potentially compatible pairs (NPCP) of qualitative characters from a data set may be considered to be a measure of its phylogenetic randomness. The value of NPCP depends on the number of evolutionary units (EUs), the number of characters, the number of states in the characters, the distributions of EUs among these states, and the amount and distribution of missing information and so does not directly indicate degree of phylogenetic randomness. Thus, for an observed data set, we used Monte Carlo methods to estimate the probability that a data set chosen equiprobably from among those identical (with respect to all the other above determining features) to the observed data set would have as high (or low) an NPCP as the observed data set. This probability, the realized significance of the observed NPCP, is attractive as an indication of phylogenetic randomness because it does not require the assumptions made by other such methods: No character state trees are assumed and consequently, only potential compatibility can be determined; no particular method of phylogenetic estimation is assumed; and no phylogenetic trees are constructed. We determined the values and significances of NPCP for analyses of 57 data sets taken from 53 published sources. All data sets from 37 of those sources exhibited realized significances of < 0.01, indicating high levels of phylogenetic nonrandomness. From each of the remaining 16 sources, at least one data set was more phylogenetically random. Inclusion of outgroups changed significance in some cases, but not always in the same direction. Data sets with significantly low NPCP may be consistent with an ancient hybrid origin (or other ancient polyphyletic gene exchange, crossing over, viral transfer, etc.) of the study group.     

Uninode coding vs gene tree parsimony for phylogenetic reconstruction using duplicate genes

Two different methods of using paralogous genes for phylogenetic inference have been proposed: reconciled trees (or gene tree parsimony) and uninode coding. Gene tree parsimony suffers from 10 serious problems, including differential weighting of nucleotide and gap characters, undersampling which can be misinterpreted as synapomorphy, all of the characters not being allowed to interact, and conflict between gene trees being given equal weight, regardless of branch support. These problems are largely avoided by using uninode coding. The uninode coding method is elaborated to address multiple gene duplications within a single gene tree family and handle problems caused by lack of gene tree resolution. An example of vertebrate phylogeny inferred from nine genes is reanalyzed using uninode coding. We suggest that uninode coding be used instead of gene tree parsimony for phylogenetic inference from paralogous genes.     

Phylogenetic interpretation of ontogenetic change: sorting out the actual and artefactual in an empirical case study of centrarchid fishes

Hypothesized relationships between ontogenetic and phylogenetic change in morphological characters were empirically tested in centrarchid fishes by comparing observed patterns of character development with patterns of character evolution as inferred from a representative phylogenetic hypothesis. This phylogeny was based on 56–61 morphological characters that were polarized by outgroup comparison. Through these comparisons, evolutionary changes in character ontogeny were categorized in one of eight classes (terminal addition, terminal deletion, terminal substitution, non-terminal addition, non-terminal deletion, non-terminal substitution, ontogenetic reversal and substitution). The relative frequencies of each of these classes provided an empirical basis from which assumptions underlying hypothesized relationships between ontogeny and phylogeny were tested. In order to test hypothesized relationships between ontogeny and phylogeny that involve assumptions about the relative frequencies of terminal change (e.g. the use of ontogeny as a homology criterion), two additional phylogenies were generated in which terminal addition and terminal deletion were maximized and minimized for all characters. Character state change interpreted from these phylogenies thus represents the maxima and minima of the frequency range of terminal addition and terminal deletion for the 8.7 × 10³⁶ trees possible for centrarchids. It was found for these data that terminal change accounts for c. 75% of the character state change. This suggests either that early ontogeny is conserved in evolution or that interpretation and classification of evolutionary changes in ontogeny is biased in part by the way that characters are recognized, delimited and coded. It was found that ontogenetic interpretation is influenced by two levels of homology decision: an initial decision involving delimitation of the character (the ontogenetic sequence), and the subsequent recognition of homologous components of developmental sequences. Recognition of phylogenetic homology among individual components of developmental sequences is necessary for interpretation of evolutionary changes in ontogeny as either terminal or non-terminal. If development is the primary criterion applied in recognizing individual homologies among parts of ontogenetic sequences, the only possible interpretation of phylogenetic differences is that of terminal change. If homologies of the components cannot be ascertained, recognition of the homology of the developmental sequence as a whole will result in the interpretation of evolutionary differences as substitutions. Particularly when the objective of a study is to discover how ontogeny has evolved, criteria in addition to ontogeny must be used to recognize homology. Interpretation is also dependent upon delimitation within an ontogenetic sequence. This is in part a function of the way that an investigator ‘sees’ and codes characters. Binary and multistate characters influence interpretation differently and predictably. The use of ontogeny for determining phylogenetic polarity as previously proposed rests on the assumptions that ancestral ontogenies are conserved and that character evolution occurs predominantly through terminal addition. It was found for these data that terminal addition may comprise a maximum of 51.9% of the total character state change. It is concluded that the ontogenetic criterion is not a reliable indicator of phylogenetic polarity. Process and pattern data are collected simultaneously by those engaged in comparative morphological studies of development. The set of alternative explanatory processes is limited in the process of observing development. These form necessary starting points for the research of developmental biologists. Separating ‘empirical’ results from interpretational influences requires awareness of potential biases in the course of character selection, coding and interpretation. Consideration of the interpretational problems involved in identifying and classifying phylogenetic changes in ontogeny leads to a re-evaluation of the purpose, usefulness and information conveyed by the current classification system. It is recommended that alternative classification schemes be pursued.     

Character analysis in morphological phylogenetics: problems and solutions

Many aspects of morphological phylogenetics are controversial in the theoretical systematics literature and yet are often poorly explained and justified in empirical studies. In this paper, I argue that most morphological characters describe variation that is fundamentally quantitative, regardless of whether they are coded qualitatively or quantitatively by systematists. Given this view, three fundamental problems in morphological character analysis (definition, delimitation, and ordering of character states) may have a common solution: coding morphological characters as continuous quantitative traits. A new parsimony method (step-matrix gap-weighting, a modification of Thiele's approach) is proposed that allows quantitative traits to be analyzed as continuous variables. The problem of scaling or weighting quantitative characters relative to qualitative characters (and to each other) is reviewed, and three possible solutions are described. The new coding method is applied to data from hoplocercid lizards, and the results show the sensitivity of phylogenetic conclusions to different scaling methods. Although some authors reject the use of continuous, overlapping, quantitative characters in phylogenetic analysis, quantitative data from hoplocercid lizards that are coded using the new approach contain significant phylogenetic structure and exhibit levels of homoplasy similar to those seen in data that are coded qualitatively.     

Character-state space versus rate of evolution in phylogenetic inference

With only four alternative character states, parallelisms and reversals are expected to occur frequently when using nucleotide characters for phylogenetic inference. Greater available character‐state space has been described as one of the advantages of third codon positions relative to first and second codon positions, as well as amino acids relative to nucleotides. We used simulations to quantify how character‐state space and rate of evolution relate to one another, and how this relationship is affected by differences in: tree topology, branch lengths, rate heterogeneity among sites, probability of change among states, and frequency of character states. Specifically, we examined how inferred tree lengths, consistency and retention indices, and accuracy of phylogenetic inference are affected. Our results indicate that the relatively small increases in the character‐state space evident in empirical data matrices can provide enormous benefits for the accuracy of phylogenetic inference. This advantage may become more pronounced with unequal probabilities of change among states. Although increased character‐state space greatly improved the accuracy of topology inference, improvements in the estimation of the correct tree length were less apparent. Accuracy and inferred tree length improved most when character‐state space increased initially; further increases provided more modest improvements. © The Willi Hennig Society 2004.     

Stability of classification of filamentous fungi under changes in character coding strategy

The characteristics of a number of filamentous fungal cultures were obtained from two previously published numerical taxonomic studies on Penicillium and Phoma. The coding strategies for some of the physiological and morphological properties employed in the original studies were re-examined and the data was re-coded by combining sets of characters into single ordered multistate characters. The different coding procedures were compared by generating average linkage (UPGMA) dendrograms which were in turn compared by calculating correlation coefficients between the final similarity matrices implied by these dendrograms. The character conversions had no significant effect on the final outcome of the clustering.     

Systematics of New World Monkeys (Platyrrhini, Primates) Based on 16S Mitochondrial DNA Sequences: A Comparative Analysis of Different Weighting Methods in Cladistic Analysis

In order to investigate the effects of different weighting methods on a phylogeny reconstruction based on DNA sequences and to evaluate the phylogenetic information content of various secondary structures, a fragment of the large ribosomal mitochondrial gene (16S) was sequenced from 13 species of New World monkeys, three species of catarrhines, and Tarsius. The data were analyzed cladistically without weighting characters or changes, and with different weighting methods: a priori differential weights for transitions and transversions, two variants of dynamic weighting for each kind and direction of change, and successive approximations, using both the character consistency index (CI) and the rescaled consistency index (RC). The results were compared with published trees constructed from nuclear sequences of ε-globins and morphological characters by different authors. The result of the analysis of the mtDNA data set with successive approximations, using the RC as weighting function, was the closest to the topology on which all molecular and morphological trees concur. Other relationships were unique to this tree. "Loops" were the type of secondary structure that showed maximum variation in sequence length and sites with the lowest character CI and RC. A large number of sites within loops showed high values for these indices, however, which suggests that uniform downweighting of these regions represents a large loss of phylogenetic information. Successive weighting, which assigns a weight for each particular character, seems to be a desirable alternative to this practice. We propose a new variant of dynamic weighting, which we call homoplasy-correcting dynamic weighting, that like dynamic weighting, is applicable to any kind of sequence, coding or noncoding.     

Morphological character evolution in hepialid moths (Lepidoptera: Hepialidae) from New Zealand

Mapping morphological characters on a molecular-based phytogeny enabled examination of character evolution and an historical perspective into evolutionary processes, both of which are important aspects of systematic research and comparative biology. In this study, 63 morphological characters from hepialid moths in New Zealand were mapped on a phylogenetic tree reconstructed from mitochondrial DNA COI & II sequence data. Morphological characters hypothesized to be synapomorphies for the New Zealand 'Oxycanus' lineages and 'Oxycanus' lineage s.s. were confirmed to be homologous when mapped on the COI & II phytogeny. The direction of character state transformation was determined for five characters, with members of the Aenetus and Aoraia lineages exhibiting hypothesized ancestral states. Male genitalic characters were less homoplasious than other character partitions and covaried significandy with phytogeny.     

THE NATURE OF CLADISTIC DATA

Abstract— Cladistic data are the characters of organisms. Character is defined as a feature that can be evaluated as a variable with two or more mutually exclusive and ordered states. Cladistic characters must be treated as multistate variables, and coded as sequential numbers or in additive binary fashion. Any other interpretation and handling of cladistic data will introduce error into analysis. Character states cannot be treated independently as present or absent, i.e., as nominal variables, because redundancy is introduced into the data and information content is sacrificed. Non-additive binary coding demonstrates that treating cladistic variables as nominal data will lead to multiple, equally parsimonious solutions. Defining characters found universally in a group of organisms, but unknown outside those organisms have no alternative state that can be designated as absent. Absence, however, is valid as a character state if it can be shown to be apomorphic. When two or more character states occur within a taxon, that taxon must be coded as having an unknown state for that character, or the taxon must be split in two or more taxa. Continuously varying quantitative data are not suitable for cladistic analysis because there is no justifiable basis for recognizing discrete states among them. Quantitative data are questionable even when they exhibit mutually exclusive states because the states can be interpreted only in reference to an archetype, i.e., as implied homologies not subject to test.     

Computerized Analysis of Crustacean Relationships

The strengths and weaknesses of phylogenetic analysis using computers are reviewed from the viewpoint of understanding crustacean evolution. Computerized methods require the explicit presentation of characters and character state homologies. New techniques allow investigators to design evolutionary models into a character data matrix, or to use evolutionary models that make minimal a priori assumptions. The computer analysis relieves the investigator from the highly repetitious testing of trees, allows the concentration on the character state data, and provides objective methods for comparing trees, primarily their length. These are regarded as the strengths of computerized methods. The weaknesses of these methods include the relatively inscrutable nature of the character data matrix compared with the overall ‘gestalt’ of resulting trees, the difficulties of defining discrete homologies within the Crustacea, especially for counts of segmentation, the lack of clear intermediate character states in some multistate segmental characters, and the inability to define evolutionary polarity. These difficulties may be overcome by analysing the data using the minimal assumption models of character evolution, and by a recognition that the trees are a result of the input data, and therefore the data should be criticized, rather than the trees themselves. A ‘consensus’ character data set, including most extant major groups of the Crustacea as well as several key fossils, was assembled and revised by the participants in the workshop. An artificial taxon, ‘ur-crustacean characters’, was introduced to root the tree. Three observations may be made from parsimony analyses using several weighting and tree rooting methods. (1) The currently accepted large scale phylogeny and classification of the Crustacea is not corroborated. (2) The number of supposed plesiomorphic traits possessed by a taxon is not a good index for early derivation in crustacean evolution. (3) The taxon Maxillopoda is not supported by the arrangement of any of the trees.     

Cladistic analysis of molecular and morphological data

Considerable progress has been made recently in phylogenetic reconstruction in a number of groups of organisms. This progress coincides with two major advances in systematics: new sources have been found for potentially informative characters (i. e., molecular data) and (more importantly) new approaches have been developed for extracting historical information from old or new characters (i. e., Hennigian phylogenetic systematics or cladistics). The basic assumptions of cladistics (the existence and splitting of lineages marked by discrete, heritable, and independent characters, transformation of which occurs at a rate slower than divergence of lineages) are discussed and defended. Molecular characters are potentially greater in quantity than (and usually independent of) more traditional morphological characters, yet their great simplicity (i. e., fewer potential character states; problems with determining homology), and difficulty of sufficient sampling (particularly from fossils) can lead to special difficulties. Expectations of the phylogenetic behavior of different types of data are investigated from a theoretical standpoint, based primarily on variation in the central parameter λ (branch length in terms of expected number of character changes per segment of a tree), which also leads to possibilities for character and character state weighting. Also considered are prospects for representing diverse yet clearly monophyletic clades in larger-scale cladistic analyses, e. g., the exemplar method vs. “compartmentalization” (a new approach involving substituting an inferred “archetype” for a large clade accepted as monophyletic based on previous analyses). It is concluded that parsimony is to be preferred for synthetic, “total evidence” analyses because it appears to be a robust method, is applicable to all types of data, and has an explicit and interpretable evolutionary basis. © 1994 Wiley-Liss, Inc.     

Coding characters from different life stages for phylogenetic reconstruction: a case study on dragonfly adults and larvae,including a description of the larval head anatomy of Epiophlebia superstes (Odonata: Epiophlebiidae)

The exclusive use of characters coding for specific life stages may bias tree reconstruction. If characters from several life stages are coded, the type of coding becomes important. Here, we simulate the influence on tree reconstruction of morphological characters of Odonata larvae incorporated into a data matrix based on the adult body under different coding schemes. For testing purposes, our analysis is focused on a well‐supported hypothesis: the relationships of the suborders Zygoptera, ‘Anisozygoptera’, and Anisoptera. We studied the cephalic morphology of Epiophlebia, a key taxon among Odonata, and compared it with representatives of Zygoptera and Anisoptera in order to complement the data matrix. Odonate larvae are characterized by a peculiar morphology, such as the specific head form, mouthpart configuration, ridge configuration, cephalic musculature, and leg and gill morphology. Four coding strategies were used to incorporate the larval data: artificial coding (AC), treating larvae as independent terminal taxa; non‐multistate coding (NMC), preferring the adult life stage; multistate coding (MC); and coding larval and adult characters separately (SC) within the same taxon. As expected, larvae are ‘monophyletic’ in the AC strategy, but with anisopteran and zygopteran larvae as sister groups. Excluding larvae in the NMC approach leads to strong support for both monophyletic Odonata and Epiprocta, whereas MC erodes phylogenetic signal completely. This is an obvious result of the larval morphology leading to many multistate characters. SC results in the strongest support for Odonata, and Epiprocta receives the same support as with NMC. Our results show the deleterious effects of larval morphology on tree reconstruction when multistate coding is applied. Coding larval characters separately is still the best approach in a phylogenetic framework. © 2015 The Linnean Society of London