Evolutionary systematists use the achieved level of anagenesis as the main guideline in defining taxa. If some substantial change in phenotype properties, an important evolutionary innovation, occurred in a certain phylogenetic line, evolutionary systematists frequently consider it useful to classify this line into a discrete taxon, separate from the other lines (Fig. XXV.4). A decision on whether a particular phenotype change is a sufficiently large innovation for its carriers to deserve to be a separate taxon remains a subjective matter. This means that delimitation of the individual taxa in a natural system is, to a considerable degree, a matter of the subjective decision of taxonomists and subsequently of convention. If he is to strictly respect the requirement on monophyly of the created taxa, a systematist must frequently also include in the particular taxon species or groups of species that do not exhibit the given, from our viewpoint key, property. Some species could have retained the original plesiomorphic form of the particular trait or could even return to this form (Fig. XXV.5). A further complication is entailed in the very real possibility that the evolutionary innovation that is to form the basis for defining the taxon could appear independently several times in a phylogenetic line and that this does not correspond to homology but rather homoplasy. It is prohibited to define a taxon on the basis of shared homoplasy; however, it is not clear whether this can always be completely avoided. In some cases, a particular trait is formed on the basis of some other trait functioning as preadaptation for its formation. Consequently, the relevant new features occur within the particular line independently in a number of species whose common ancestor did not exhibit the particular new trait (Fig. XXV.6). In this case, a solution could lie in defining the particular taxon on the basis of the presence of a certain preadaptation; however, it is primarily necessary to consider whether the defining of a taxon on the basis of this trait is at all useful.