438 Brian K. Hall
of life, however, taxa possessing convergent features share an evolutionary his-
tory, a history that includes conserved regulator (developmental) genes, gene net-
works/cascades and epigenetic processes such as cell aggregation, migration, dif-
ferentiation and morphogenesis.
Darwin saw the difficulties of separating convergence from analogy. Determina-
tion of relationships between major groups of organisms and the search during the
19 thC for the group that gave rise to the chordates/vertebrates was complicated
enormously by the inability to determine whether segmentation had evolved once
(i.e., was homologous) or more than once and so was convergent. The same was
(and is still) true for the origination of mesoderm(s).^18
Arthur Willey, best known for his studies on amphioxus (Branchiostoma), and
motivated by the plethora of theories of the origin of the vertebrates, devoted
a book to convergence [Willey, 1911].^19 As with naming a new species ofHomo
rather than assigning your find to an already named taxon, Willey noted that there
was “more joy amongst morphologists over one attempt at genealogy than over
ninety and nine demonstrations of convergence” [1911, 53]. Willey did not confine
himself to listing cases of convergence but proposed “that homoplasy does not
cover all the cases which are included under convergence in the wider acceptance
of the term” [and that] “all homoplasy is convergence, but all convergence is
not homoplasy” [1911, xii, 11]; Willey saw that phylogenetic relations had to be
established before convergence could be considered.
9 PARALLELISMParallelism, which has been described as a “gray zone between homology and con-
vergence” [Gould, 2002, 1088], describes the presence (or formation) of a feature
in related lineages but not in their most recent common ancestor and/or sister
group If the feature were present in the most recent common ancestor it would
be a homologue. If the feature was present in a more distant lineage (but not in
the most recent common ancestor, its reappearance would be convergence (above).
Loss and reappearance of a feature in a lineage is reversal (see below). Clearly
knowledge of phylogenetic relationships is critical to our ability to separate ho-
mologous, convergent and parallel features. The expectation is that “if [character
states] are parallelisms, then they should be developmentally and genetically the
same” [Kellogg and Shaffer, 1993, 412].
Simpson [1961, 78] defined parallelism as “the development of similar characters
separately in two or more lineages of common ancestry and on the basis of, or
channeled by, characteristics of that ancestry”. The phrase “characteristics of
(^18) See Ospovat [1995, 111–112] for Darwin on convergence and analogy, and see Bowler [1996],
Freeman and Martindale [2002] and Martindaleet al.[2004] for the origin(s) of mesoderm(s).
(^19) Willey’s is perhaps the only book devoted to convergence, although Gregory [1951] lists well
over 100 examples of convergence, and the treatises on variation and variability by Bateson [1894]
and Yablokov [1974] include examples of convergence. See Bowler [1996] for Willey’s views on
parallelism and convergence.