Dispelling five myths about hypothesis testing in biological systematics

The emphasis on testing phylogenetic hypotheses has been prominent since the English language introductions of Willi Hennig’s Phylogenetic Systematics (1966) and Karl Popper’s (1959) Logic of Scientific Discovery. While the mechanics of hypothesis and theory testing are well established in other fields of science, adherence to those prescriptions in biological systematics has rarely been formally recognized. The consequence has been the development of potentially contradictory approaches to empirically evaluating phylogenetic hypotheses under the guise of testing. In his brief review of the topic, Assis (Cladistics, 30:240–242, 2014) identified five different views on phylogenetic hypothesis testing: (1) total evidence, (2) taxonomic congruence, (3) reciprocal illumination, (4) homology assessment, and (5) taxa sampling. The present paper examines the validity of these views against the actual inferential steps required to infer hypotheses and subsequently engage in testing, i.e., abduction, deduction, and induction (sensu stricto), respectively. It is shown that none of the tests discussed by Assis are valid, and while it is straightforward to outline what is required to properly test phylogenetic hypotheses, the feasibility of accomplishing such tests is operationally impractical in most instances.