Portal de Periódicos da CAPES

Carta Produção Nacional Revisado por pares

BIBTEX RIS

Testing evolutionary hypotheses: from Willi Hennig to Angiosperm Phylogeny Group

2013; Wiley; Volume: 30; Issue: 3 Linguagem: Inglês

10.1111/cla.12048

1096-0031

Leandro C. S. Assis,

Plant and animal studies

The person who requires a phylogenetic system as a premise for his own work, will then have to decide on which side lie the better arguments; the criteria for this must again emerge from the theory of phylogenetic systematics. Differences of opinion on matters of fact are not, however, a special defect of phylogenetic systematics but the universal mark of every science. (Hennig, 1965, p. 115) Sir, Regarding a recent, revived debate in Cladistics on phylogeny and hypothesis testing (Vogt, 2008, 2013; Farris, 2013a,b; and references therein), I present in this letter an additional point of view in honour of Willi Hennig's legacy in comparative biology (see also Wheeler et al., 2013). My main point is to discuss hypothesis formulation and testing with respect to (i) total evidence, (ii) taxonomic congruence, (iii) reciprocal illumination, (iv) homology assessment, and (v) taxa sampling. To complement this, the Angiosperm Phylogeny Group (APG) is treated as a study case. The debate concerning the role of morphological and molecular evidence is also revisited. To conclude, some lessons from Hennig are remembered. I hope the issues briefly addressed here further stimulate the development of a dynamic and healthy background with regard to hypothesis formulation and testing in comparative biology. With the development of Hennig's (1966) Phylogenetic Systematics, hypotheses of evolutionary relationships could be categorically formulated and tested. Rooted in the Hennigian plan of argumentation, and with the implementation of sophisticated operations, cladistic theory and practice emerged as a successful approach (Schuh and Brower, 2009). More recently, model-based methods have begun to be employed (Ronquist et al., 2009; Schmidt and von Haeseler, 2009). Once a hypothesis of evolutionary relationship is formulated (i.e. from the data matrix to the phylogenetic tree), there are five approaches to testing it: Researchers formulating and testing evolutionary hypotheses could make it explicit which of these approaches they would employ, as well as justify the employment itself. Although evolutionary hypotheses can be tested they cannot be falsified, as they are historical hypotheses about singular events, and, in fact, we are unable to come back to the past to check whether they are true or false (Vogt, 2008, 2013; and references therein). In some areas of comparative biology, the testing of evolutionary hypotheses seems to be misunderstood. In systematic botany, for instance, some researchers have usually accepted the most current hypothesis, such that previous hypotheses seem to be considered either false or obsolete. As a study case, the APG has updated, from 1998 to 2009, its influential system of classification based on phylogenetic hypotheses. During this period, the group has provided three classificatory schemata (APG I, 1998; APG II, 2003; APG III, 2009), from the level of family to higher groups. Its system of classification has been dogmatically communed by botanical science, even reaching other areas such as ecology (e.g. Webb and Donoghue, 2005). Indeed, researchers working with angiosperm groups have usually cited the most recent APG classification11 The journal Phytotaxa, for instance, informs authors: "[…] it is strongly recommended that family classification follows […] APG III (2009), see also Reveal & Chase (2011) […]. The use of alternative family concepts will require a written justification." (http://www.mapress.com/phytotaxa/author.htm) and seem to have forgotten that previous hypotheses concerning the phylogeny of angiosperms (e.g. Donoghue and Doyle, 1989; Doyle et al., 1994; Doyle and Endress, 2000) cannot be falsified and are therefore valid. The validity of all these hypotheses, as well as their use, makes systematic botany more scientifically dynamic and wealthy, rather than using APG as a universal system of reference. Another interesting point is that these latter hypotheses (Donoghue and Doyle, 1989; Doyle et al., 1994; Doyle and Endress, 2000) were supported by cladistic analyses of morphological data, whereas the APG system is based on a summary figure of several published hypotheses predominantly supported by chloroplast DNA sequences. The way APG deals with hypotheses is not contemplated in the four approaches detailed above. Nevertheless, one could object by saying that APG contemplates taxonomic congruence. More aptly but not explicitly, APG deals with taxonomic congruence sensu Miyamoto (1985), in which a consensus tree, based on topologies resulting from distinct data matrices and methods, constitutes the final tree figure. By contrast, taxonomic congruence (modified from Mickevich, 1978) is not a summary of congruent hypotheses in a new final diagram, but rather confronts (tests) hypotheses resulting from distinct evidence to identify why conflicts appear. If APG had used the taxonomic congruence (sensu Miyamoto, 1985) of all published phylogenetic hypotheses addressed to the level of family and higher groups, the result would potentially be a poorly resolved consensus tree (including monocots and eudicots), as there exist conflicting hypotheses relative to the angiosperm tree of life. If this makes sense, arguments for selecting hypotheses used to draw the final tree deserve further clarifications. In the past two decades, phylogenetic analyses based on morphological characters have been rarely conducted (Bybee et al., 2009). In turn, the role of morphology in phylogenetics has been poorly reduced to character mapping onto DNA-based trees. After mapping, normal practice is to point out either a single or multiple character origins. Following this approach, mapped phenotypes are not, but should be (i.e. reciprocal illumination), treated as relevant evidence to test DNA-based evolutionary hypotheses. They are viewed as "mere passengers" (Maddison, 2006, p. 1743) onto the trees supported by DNA regions (e.g. plastidial, mitochondrial, ITS) that have no causal association with their developmental evolution or homology (Assis and Santos, 2013). Analyses of morphological characters (and other kinds of evidence) are now needed to test DNA-based hypotheses, using total evidence, taxonomic congruence, reciprocal illumination, homology assessment, and taxon sampling (cf. Bateman et al., 2006; Rieppel, 2007; Gauthier et al., 2012). In squamates, for instance, the multiple origins of complex phenotypes in the context of molecular phylogenetic analysis have been discounted (tested) using taxonomic congruence in the context of morphological phylogenetic analysis (Gauthier et al., 2012). It is also relevant to highlight that morphological characters may be continuously revised throughout comparative analyses (i.e. homology assessment) (Rieppel, 2007), thus testing evolutionary hypotheses. In the case of DNA sequences alignment, however, this practice is rare, but not insignificant (Wägele et al., 2009). For comparative biology, fundamental lessons may be learned from Willi Hennig. Based on the critical analysis of homology, Hennig endorsed the use of different kinds of evidence to discover evolutionary relationships, as well as comparing evolutionary hypotheses resulting from different ontogenetic stages (i.e. semaphoronts) in a search for congruent or incongruent groupings. In addition, he proposed the analysis of reciprocal illumination to test whether the hypotheses make sense in light of additional, relevant information. All these approaches continue to be essential for hypothesis formulation and testing in comparative biology. Hennig's lessons should not be forgotten (Wheeler et al., 2013). I thank Leonardo Borges, Mariana Bünger, Peter Endress, Luiza Paula, Olivier Rieppel, Leandro Santos, João Renato Stehmann, and the Editor Dennis Stevenson for valuable comments on an early draft of this letter.