The 'natural system' of organisms reflects their phylogenetic relation
ship. It is the result of an historical process and has to be inferred
from the available evidence. In the morphological phenotype, historic
al traces are intermeshed with functional adaptations. Overall similar
ity, even if quantified, can be a misleading indicator of relatedness.
Cladistics uses shared derived character states (synapomorphies) to i
dentify groups of common ancestry. Synapomorphies are mostly inferred
from their taxonomic context. If apparently equally valid characters s
uggest mutually exclusive groups, parsimony is invoked: a phylogenetic
reconstruction requiring a minimum of evolutionary steps to describe
the present character distribution is accepted as the most likely one.
Cladistics sets very stringent requirements for informative character
s, and a rigorous analysis of morphology is likely to yield very few r
eliable characters. The direct analysis of DNA sequences provides theo
retically the optimal evidence for phylogenetic reconstruction. In pra
ctice, very little of this information is readily accessible. Occasion
ally major sequence rearrangements can be unequivocal synapomorphies.
Many phylogenetic problems can be solved by comparative sequencing of
an appropriate segment of DNA. Comparative sequencing of the chloropla
st gene rbcL has become the model for such studies. Molecular data hav
e confirmed much traditional taxonomy, elucidated doubtful cases and c
orrected misinterpretations. Molecular data also have clearly shown th
e limits of the cladistic approach by revealing both known and previou
sly unsuspected modes of reticulate evolution. Molecular approaches se
parate phylogenetic reconstruction from biological evaluation and will
never replace morphological analysis in Systematics. However, molecul
ar methods also facilitate the direct investigation of morphological e
volution by revealing the genetic basis of morphogenesis in model syst
ems or by permitting the genetic analysis of diagnostic character chan
ges by genetic mapping.