Arnold G. Kluge

PARSIMONY JACKKNIFING OUTPERFORMS NEIGHBOR-JOINING

Abstract— Because they are designed to produced just one tree, neighbor‐joining programs can obscure ambiguities in data. Ambiguities can be uncovered by resampling, but existing neighbor‐joining programs may give misleading bootstrap frequencies because they do not suppress zero‐length branches and/or are sensitive to the order of terminals in the data. A new procedure, parsimony jackknifing, overcomes these problems while running hundreds of times faster than existing programs for neighbor‐joining bootstrapping. For analysis of large matrices, parsimony jackknifing is hundreds of thousands of times faster than extensive branch‐swapping, yet is better able to screen out poorly‐supported groups.

CLADISTICS: WHAT'S IN A WORD?

Abstract— Cladistics has changed considerably with the availability of new methods and sources of data, and the increasing realization that cladograms are relevant to all manner of historical questions. Criticisms of, and justifications for, consensus hypotheses in phylogenetic inference are reviewed. The conclusion is overwhelmingly against taxonomic congruence which deliberately seeks consensus propositions. The total evidence approach is not so burdened. A preference for suboptimal cladograms is also critized, as is the protocol for mapping characters of special interest onto a phylogenetic hypothesis derived from other evidence. The bootstrap and jackknife resampling techniques are questioned because their underlying assumptions are violated and they are sensitive to character frequencies. These findings suggest that cladistics is being redefined in ways that contradict the practices and principles responsible for its pre‐eminence in phylogenetic inference.

SKEWNESS AND PERMUTATION

Abstract— The skewness criterion of phylogenetic structure in data is too sensitive to character state frequencies, is not sensitive enough to number of characters (degree of corroboration) and relies on counts of arbitrarily‐resolved bifurcating trees. For these reasons it can give misleading results. Permutation tests lack those drawbacks and can be performed quickly by using approximate parsimony calculations, but the test based on minimal tree length can imply strong structure in ambiguous data. A more satisfactory test is obtained by using a support measure which takes multiple trees into account.

Testability and the Refutation and Corroboration of Cladistic Hypotheses

Both refutationist and verificationist philosophies and practices are becoming increasingly evident in phylogenetic inference. Refutation and verification are fundamentally different epistemologies, and it seems unlikely that they can continue to coexist as the basis for inferring species history. The refutationist nature of cladistics is explored in terms of Popperian testability, in order to understand better the alternatives and to predict the outcome of the expected revolution. Testability concerns the logical relationship between a hypothesis (h, such as a cladogram), evidence (e, such as synapomorphy), and background knowledge (b). Of particular importance is the direct relationship between the logical improbability ofhand its potential to be tested, because foreto corroborateh,eshould be improbable givenbalone. Simplicity and boldness, amount of empirical content, and logical improbability all refer to the potential to be tested. Thathmust be testable by severe tests is the same as saying that those tests have greater probability of failing, given onlyb. Descent with modification is sufficient as background knowledge (b) in phylogenetic inference, and such a minimal assumption explains the generality of cladistics. Also of interest to the refutationist position is total evidence. In terms of testability, a statement describing the results of multiple tests is less probable than a statement describing only some of the tests, the multiple test result being more improbable, and accordingly more severe, than its component tests. All other cladistic principles and practices considered in this review are also understandable in terms of Popperian testability, refutation and corroboration. These include minimizing ad hoc hypotheses of homoplasy and minimizing explanatory power, and choosing tentatively among cladograms according to their degree of corroboration (support). Differential character weighting is determined to be unacceptable in terms of testability. Also, testability does not provide a basis for assessing the accuracy of hypotheses, but then that is of no consequence to cladists, because they are not preoccupied with knowing the absolute truth, unlike verificationists.

Data exploration in phylogenetic inference: scientific, heuristic, or neither

The methods of data exploration have become the centerpiece of phylogenetic inference, but without the scientific importance of those methods having been identified. We examine in some detail the procedures and justifications of Wheelers sensitivity analysis and relative rate comparison (saturation analysis). In addition, we review methods designed to explore evidential decisiveness, clade stability, transformation series additivity, methodological concordance, sensitivity to prior probabilities (Bayesian analysis), skewness, computer‐intensive tests, long‐branch attraction, model assumptions (likelihood ratio test), sensitivity to amount of data, polymorphism, clade concordance index, character compatibility, partitioned analysis, spectral analysis, relative apparent synapomorphy analysis, and congruence with a “known” phylogeny. In our review, we consider a method to be scientific if it performs empirical tests, i.e., if it applies empirical data that could potentially refute the hypothesis of interest. Methods that do not perform tests, and therefore are not scientific, may nonetheless be heuristic in the scientific enterprise if they point to more weakly or ambiguously corroborated hypotheses, such propositions being more easily refuted than those that have been more severely tested and are more strongly corroborated. Based on common usage, data exploration in phylogenetics is accomplished by any method that performs sensitivity or quality analysis. Sensitivity analysis evaluates the responsiveness of results to variation or errors in parameter values and assumptions. Sensitivity analysis is generally interpreted as providing a measure of support, where conclusions that are insensitive (robust, stable) to perturbations are judged to be accurate, probable, or reliable. As an alternative to that verificationist concept, we define support objectively as the degree to which critical evidence refutes competing hypotheses. As such, degree of support is secondary to the scientific optimality criterion of maximizing explanatory power. Quality analyses purport to distinguish good, reliable, accurate data from bad, misleading, erroneous data, thereby assessing the ability of data to indicate the true phylogeny. Only the quality analysis of character compatibility can be judged scientific—and a weak test at that compared to character congruence. Methods judged to be heuristic include Bremer support, long‐branch extraction, and safe taxonomic reduction, and we underscore the great heuristic potential of a posteriori analysis of patterns of transformations on the total‐evidence cladogram. However, of the more than 20 kinds of data exploration methods evaluated, the vast majority is neither scientific nor heuristic. Given so little demonstrated cognitive worth, we conclude that undue emphasis has been placed on data exploration in phylogenetic inference, and we urge phylogeneticists to consider more carefully the relevance of the methods that they employ.

Total Evidence Or Taxonomic Congruence: Cladistics Or Consensus Classification

Miyamoto and Fitchs (1995,Syst. Biol. 44: 64–76) verificationist arguments for taxonomic congruence are evaluated and found to be unconvincing. In particular, there is no logical connection between the truth of phylogenetic hypotheses and the independence of the sets of characters analysed for their consensus. Further, the character set partitions emphasized by Miyamoto and Fitch must be considered arbitrary, because they are based on untestable process assumptions.

Probabilism and Phylogenetic Inference

The maximum likelihood approach to phylogenetics rests on frequency probability theory. This stands in stark contrast to the logical probability of corroboration‐based cladistic parsimony. History is particular and cannot be described in terms of universal statements about abstract generalities, the task of the historical sciences being one of explanation, not prediction. Thus, frequency probability methods of estimation are inappropriate for making historical inferences. Maximum likelihood estimation procedures are deconstructed from numerous perspectives in spite of their supposed impressive technicalities. Charges of parsimonys inconsistency are rendered mute, because its justification lies elsewhere, yet maximum likelihood is still subject to Walds dilemma if realism is of any interest. Although all epistemologies make assumptions, the models employed by maximum likelihood are problematic and deterministic, as opposed to the unproblematic background knowledge characteristic of cladistics. Apart from issues of logical and sampling dependencies, the requirements of frequency probability theory are non‐trivial and the maximum likelihood estimation of phylogeny can neither escape, nor satisfy the tenets of calculus independence (e.g. i.i.d.) inherent in the multiplicative relations of the method. If phylogeneticists are to maintain a rational foundation for their epistemology, neo‐justificationist appeals to some metaphysical truth must be abandoned in favour of the realism of sophisticated falsification.

Sophisticated falsification and research cycles: Consequences for differential character weighting in phylogenetic systematics

Practicing phylogenetic systematics as a sophisticated falsification research program provides a basis for claiming increased knowledge of sister species relationships and synapomorphies as evidence for those cladistic propositions. Research in phylogenetic systematics is necessarily cyclic, and the place where the positive shift in understanding occurs is subsequent to discovering the most parsimonious cladogram(s). A priori differential character weighting is inconsistent with seeking the maximally corroborated cladogram (sensu Popper), because weighting adds to background knowledge, the evidence being then less improbable than it would be otherwise. Also, estimating weights from character state frequencies on a cladogram is inconsistent with the view that history is unique. Sophisticated falsification provides the place in the cycle of phylogenetic systematic research where weight of evidence can be evaluated and these inconsistencies do not apply. On balance, phylogenetic systematics appears to achieve greater coherence and generality as a result of focusing on the foundations for claiming increased knowledge and avoiding efforts to differentially weight characters.

Transformation Series as an Ideographic Character Concept

An ideographic concept of character is indispensable to phylogenetic inference. Hennig proposed that characters be conceptualized as “transformation series”, a proposal that is firmly grounded in evolutionary theory and consistent with the method of inferring transformation events as evidence of phylogenetic propinquity. Nevertheless, that concept is usually overlooked or rejected in favor of others based on similarity. Here we explicate Hennigs definition of character as an ideographic concept in the science of phylogenetic systematics. As transformation series, characters are historical individuals akin to species and clades. As such, the related concept of homology refers to a historical identity relation and is not equivalent to or synonymous with synapomorphy. The distinction between primary and secondary homology is dismissed on the grounds that it conflates the concept of homology with the discovery operations used to detect instances of that concept. Although concern for character dependence is generally valid, it is often misplaced, focusing on functional or developmental correlation (both of which are irrelevant in phylogenetic systematics but may be valid in other fields) instead of the historical/transformational independence relevant to phylogenetic inference. As an ideographic science concerned with concrete objects and events (i.e. individuals), intensionally and extensionally defined properties are inconsistent with the individuation of characters for phylogenetic analysis, the utility of properties being limited to communicating results and facilitating future rounds of testing.

ONTOGENY AND PHYLOGENETIC SYSTEMATICS

Abstract— Dedifferentiation, paedomorphosis, and the insertion and deletion of developmental stages make it impossible to deduce the genealogical hierarchy from only ontogenetic transformation series. Like the outgroup criterion, ontogenetic character precedence is not theory‐neutral and to use it to deduce genealogy requires certain assumptions.