In phylogenetic analysis, when faced with different data sets from dif
ferent sources, two approaches, ''taxonomic congruence'' and ''total e
vidence'', compete with one another in seeking a best-fitting hypothes
is. In either approach one deliberately ignores the notorious uncertai
nty typically associated with the estimate of a best tree. It is propo
sed here to represent the partly conflicting information contained in
the data by specific networks with reticulations. This allows more fle
xibility in evaluating and eventually combining data.