Downloaded from Brill.Com10/09/2021 09:23:10AM Via Free Access 68 R.A
Total Page:16
File Type:pdf, Size:1020Kb
Contributions to Zoology, 71 (1/3) 67-91 (2002) SPB Academic Publishing bv, The Hague Boolean logic and character state identity: pitfalls of character coding in metazoan cladistics Ronald+A. Jenner Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Mauritskade 57, 1092AD Amsterdam, The Netherlands. Present address: University Museum of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, United Kingdom character character Boolean Keywords: metazoan cladistics, Metazoa, coding, state identity, logic, non- additive binary coding, absence/presence coding Abstract Reconstructing body plan evolution with Boolean logic: narrating history without looking back 75 When becomes congruence meaningless: unspecified A critical study ofthe morphological data sets used for the most “absence” states and uninformative character of rather recent analyses metazoan cladistics exposes a cava- reversals 77 lier attitude towards character coding. Binary absence/presence Reasserting the central role of comparative morphology is but without This coding ubiquitous, any explicitjustification. in metaz'oan cladistics 79 uncompromisingapplication ofBoolean logic in character coding ■ Conclusions 83 is remarkable since several recent investigations have nomi- Acknowledgements 84 nated absence/presence coding as the most problematic coding References 84 method available for standard cladistic analysis. Moreover, the in Appendix: unspecified “absence” states metazoan prevalence ofunspecified “absence” character states in the pub- cladistics 87 lished data sets introduces a discrepancy between the theoreti- cal foundations ofphylogenetic parsimony and current practices “For evolutionary biologists, characters transform from one in metazoan cladistics. Because phylogenetic parsimony assumes condition into another.” Kitching et at., 1998, p.25. transformation of character states, its effective operation breaks down when not all character states are carefully delimited. Ex- amples ofresulting meaningless character state transformations Introduction are discussed in two categories: 1) when unspecified “absence” states are plesiomorphic; and 2) when unspecified “absence” states To facilitate are apomorphic (character reversals). future The self-evident fact that the structure of the data progress in metazoan cladistics, the mandatory link between matrix determines the outcome of a cladistic analysis comparative morphology and character coding needs to be re- hardly needs mentioning. Data matrix construction established through a more explicit study ofmorphological vari- is also the difficult ofa cladistic ation arguably most step prior to character coding, and through a more explicitly and experimental approach to character coding. analysis, it is the only anchor that connects a cladogram to the empirical world. However, a re- markable paradox of cladistic practice in metazoan Contents phylogenetics then becomes apparent. This most important and difficult aspect of cladistic analyses Introduction 67 has received surprisingly little explicit attention, ( Some fundamentals of phylogenetic parsimony: either theoretical or practical, since the first com- primary homology, transformational homology, and puter-assisted morphological cladistic analyses of character state identity 68 the animal kingdom were published over a decade the world Boolean the Seeing through eyes: prevalence ago. This observation becomes especially striking of binary character coding in metazoan cladistics 69 when the lack of attention A/p coding and metazoan cladistics: generalproblems 71 one compares explicit The failure of Boolean logic: character state identity directed towards construction of a robust morpho- and character in unspecified “absence” states logical data set with the intensive efforts to extract metazoan cladistics 73 phylogenetic signal from a given matrix. Downloaded from Brill.com10/09/2021 09:23:10AM via free access 68 R.A. Jenner- Boolean logic and character state identity character in Several authors have noted a general trend in ing on coding the five most recently of the Metazoa that: contemporary phylogenetic research where an in- published cladistic analyses data set crease in the emphasis on the phylogenetic analy- 1) used a morphological (sometimes part of total evidence and most sis of a given data set is paralleled by a decrease a analysis), 2) sampled Giribet al. in the explicit attention directed towards construct- of the major animal taxa: et (2000) (based Sorensen ing that morphological data set (Grande & Bemis, on the data set ofZrzavy et al., 1998), et Peterson & 1998; Poe & Wiens, 2000; Rieppel & Kearney, al. (2000), Nielsen (2001), Eernisse and et al. This for the 2002). This imbalance between two necessary as- (2001), Zrzavy (2001). paper time the critical issue of whether pects ofcladistic analyses (character definition and first addresses of characters in phylogenetic analysis) is succinctly epitomized by current practice in the coding meta- theoretical the most recent comprehensive study of higher-level zoan cladistics is consistent with the molecular of arthropod relationships based upon both underpinnings phylogenetic parsimony analysis. and morphological data (Giribet et al., 2001). The authors found it worth mentioning that they per- formed 120 independent phylogenetic analyses by Some fundamentals of phylogenetic parsimony: varying sets ofparameters and datapartitions, “exe- primary homology, transformational homology, 2 cuted in parallel in the 256 processors, totaling and character state identity months of intense computation time using extremely be charac- effective tree search algorithms and an aggressive First, some terms need to defined. By 42 I of search strategy, equivalent to years of comput- ter coding mean the definition (delimitation) in sin- character its character the ing time if analyses had to be conducted a a and states, i.e., con- in gle-processor machine.” Nevertheless, not a single struction of columns a data matrix. By character cladistic character or character state transformation scoring I mean the assignment of different character of the is mentioned in their paper! Instead, the characters states to the terminal taxa, the filling in data matrix. authors are listed only in a supplementary appendix that columns of a Many designate character Both can exclusively be accessed online. both these steps as coding. steps be rather rooted in careful This imbalance may in some instances a should be morphological study. limitations in the selection of characters be understood as harmless reflection of space me- The can be character well Jenner dium chosen to report the results, as may ex- an aspect of coding as (see pected for the comprehensive study of arthropod & Schram, 1999, and especially Jenner, submitted, relationships of Giribet et al. (2001). However, in for discussions of character selection in metazoan other cases, the imbalance may signal a problem cladistics). in data it is to summarize the funda- the quality of the compiled morphological Second, necessary the mental set. Serious concerns about quality of cladistic assumptions of phylogenetic parsimony anal- data matrices traverse a broad spectrum of taxa; ysis in order to appreciate the merit of currently these include parasitic flatworms (Rohde, 1996), adopted character coding schemes in metazoan cla- hydrozoans (Marques, 1996), crustaceans (Watling, distics. At this point the reader should note that the 1999; Fryer, 1999, 2001; Olcsen, 2000), arthropods following selective references to the literature on (Shultz & Regier, 2000; Schram & Jenner, 2001), the theoretical foundations of cladistics are not reptiles (Lee, 1995; Rieppel & Reisz, 1999; Rieppel meant to be anything close to comprehensive, in- & Zaher, 2000; Rieppel & Kearney, 2002), fishes stead I focus chiefly on the most recent discussions (Patterson & Johnson, 1997; Grande & Bemis, and syntheses. and Jenner & The first in cladistic 1998), metazoans (Nielsen, 1998; very step any morphological in Schram, 1999; Jenner, 2001, submitted). This ar- analysis is a study of morphology/anatomy a ticle addresses this problem in detail for cladistic comparative framework to identify comparable fea- of the basis analyses Metazoa. tures in different taxa on the of conjectures In the I will the following, critically review cur- of similarity (De Pinna, 1991; Brower & Scha- rent practice of data matrix construction by focus- waroch, 1996; Hawkins et al., 1997; Rieppel & Downloaded from Brill.com10/09/2021 09:23:10AM via free access Contributions to - Zoology, 71 (1/3) 2002 69 This establishes Kearney, 2002). study so-called treated as such hy- by computer-assisted cladistic analy- potheses of that and primary homology are codified as ses, since cladograms are constructed and eval- characters and character in a cladistic data states uatedon the basis of character state transformations, matrix. The aim of character it coding is to represent is crucial to precisely delimit and definealternative as as observed accurately possible organismic vari- character on states the basis of careful morphological ation in a format to amenable cladistic analysis. This In other in to study. words, order represent mor- is the foundationofall computer-assisted cladistic phological variation as accurately as possible, and of analyses metazoan morphology to date in order to able to published be meaningfully interpret char- that employ phylogenetic parsimony analysis, also acter state transformations