317 



Blattai-ia and Mantodea have then been reciprocally used as outgroups, and many features 

 of the common ground-plan of Blattaria and Mantodea could be reconstructed. Then, in 

 the discussions of phallomere evolution in 7.2. and 7.3., holophyletic subgroups have been 

 established according to their hierarchy. It was begun with the search for apomorphic 

 character states common to several species, permitting the delimitation of higher-ranked 

 subgroups (subgroups 1.2., 2.1., and 2.2.). On this level, "apomorphic" relates to a 

 comparison with features well-ascertained for the common ground-plan of Blattaria and 

 Mantodea. These higher-ranked subgroups, if their holophyly could be well ascertained, 

 were then split into more subordinate subgroups, again by searching apomorphic character 

 states common to part of the species. At this level, "apomorphic" relates, if e.g. a Blattarian 

 subgroup is under consideration, to a comparison either with the common ground-plan of 

 Blattaria and Mantodea, or with the ground-plan of Blattaria, or with the ground-plan of 

 any Blattarian subgroup superordinate to and including the subgroup under consideration. 

 At last, in 7.6., the distribution of the states of the characters inconsistent with the majority 

 has been discussed in terms of parsimony. 



This hierarchical analysis has to be continuously accomplished with a procedure of 

 reciprocal illumination: There has to be a mutual feedback between the characters used, 

 also concerning their evidence in terms of phylogeny. This includes a continuous feedback 

 to the dehmitation of superordinate subgroups when working on subordinate subgroups, 

 since an autapomorphy of a superordinate subgroup might be absent within a subgroup 

 suggested to be subordinate to it, and whether a reversal has ocurred or whether the range 

 of the superordinate subgroup has to be modified by removing the subordinate subgroup 

 from it has to be discussed in terms of parsimony. 



The feedback between characters and also the resulting preliminary assumptions on 

 phylogenetic relationships can be necessary at various levels of the phylogenetic analysis: 

 for the interpretation of morphology in terms of homology relations, for the assignment 

 of a certain morphological condition present in certain species to a certain character state, 

 as well as for recognising the polarity of character states within a certain subgroup (and, 

 consequently, for the definition and formulation of characters and character states, too). 

 Hence, in the present analysis, the assumptions and conclusions related to these issues and 

 concerning certain subgroups are in many characters dependent on the distribution of 

 apomorphic states of other characters regarded as autapomorphies of a subgroup 

 superordinate to that under consideration. A character list and a matrix independent of 

 previous reciprocal illumination and preliminary assumptions on phylogeny do not include 

 this kind of feedback between characters (and their evidence) and are consequently 

 incomplete or even highly misleading in some characters. The following examples shall 

 illustrate this topic. 



(1) Concern: Interpretation of morphology in terms of homology relations. 

 As discussed in 6.3.4., the fused sclerites L3 and L4K and the muscle 14 of Ergaula 

 capucina resemble L3 and 114 of Blattellidae and Blaberidae, Anaplecta excluded. These 

 14 and 114 have been regarded as non-homologous, and the similar position of the anterior 

 insertion of the muscle moving hla - 14 or 114 - is not a synapomorphy of these taxa. 

 This hypothesis is only in part based on a homology analysis - using the criteria of relative 



