Doyle and Endress 2000). Alternatively, if these branch lengths are correct, then the
inferred rates may not be, resulting in estimates of the time intervals between cladogenic
events that are too large. The rapid morphological diversification of early angiosperms must
then have been accompanied by more, or equally, rapid molecular change, and it also
implies that rates of DNA change accelerated and slowed down in several lineages in a
correlated manner over time. This seems highly unlikely but, if it did occur, it would
appear to imply historical environmental changes as the cause (the only cause that could
simultaneously affect multiple lineages occurring worldwide would have to be global
environmental change). A pattern with both rapid morphological and molecular change
would,however, contrast with that seen in groups that have diversified more recently such
as the Asterales and Lamiales. Here, we see no apparent correlation between rapid
diversification and morphological change on the one hand, and molecular change on the
other (Soltis et al. 2000; Magallón and Sanderson 2001), and it seems illogical for more
recent patterns to be qualitatively different from older ones. Bateman (1999) tried to
address this issue of correlated or non-correlated change of morphological and molecular
characters by looking at architectural radiations on volcanic islands, and this island
approach may provide a way to address this issue at a more general level. Furthermore,
comparisons between our molecular-based estimates and those based on fossils for ‘basal’
angiosperm and eudicot lineages indicate a reasonable amount of congruence between the
two (Figures 8.3–8.4), and if we accept fossil-based estimates such as the Barremian-
Aptian split between the Nymphaeaceae and Cabombaceae (Friis et al. 2001), the
Barremian-Aptian crown-group origin of the Chloranthaceae (Friis et al. 1999), and the
early Albian split between the Platanaceae and Proteaceae (Crane and Herendeen 1996),
the crown-group origins of angiosperms and eudicots become, almost by necessity, older
than our current fossil-based estimates indicate.
Fossil-based estimates within magnolids and eudicotsIn assessing absolute diversification rates in angiosperms, Magallón et al. (1999) and
Magallón and Sanderson (2001) compiled and summarized the available fossil evidence for
the diversification times of major angiosperm lineages. We have used their summaries to
compare our molecular estimates with those based on fossil evidence. One complicating
factor is that their ages mostly refer to crown-groups, and there is no reasonable way to
deal with crown-group ages for the molecular estimates. In the molecular analyses, the
thoroughness of taxon sampling varies considerably between different clades, and this
complicates a direct comparison with their crown-group estimates. We have therefore
treated all their estimates as stem-group ages. However, for most of their crown-group
ages, they have either specified a less inclusive taxon that their estimate was based on, or
provided supporting references for their age estimates. By using either the less inclusive
taxon they specified, or referring to the original literature, we have extended our
comparisons to encompass also the less inclusive taxa. Their estimate of the
Aristolochiaceae-Lactoridaceae clade, for example, was based on the occurrence of
Lactoridaceae pollen from the Turonian of SW Africa (Zavada 1987). We have thus not
only compared their Aristolochiaceae-Lactoridaceae clade estimate with our estimate for
the split between the Aristolochiaceae-Lactoridaceae clade and its sister group, but also
ANGIOSPERM DIVERGENCE 157