papers appeared providing the skeleton of a more rigorous hierarchical framework in
which the fossil data could be interpreted (Doyle and Donoghue 1986, 1987; Donoghue
and Doyle 1989a,b; Chase et al. 1993; Doyle et al. 1994). Following this progress, a more
or less coherent picture of an Early Cretaceous origin followed by a rapid diversification
of early angiosperm lineages has appeared, and Crane et al. (1995) described the
Valanginian (135–141 Ma) appearance of angiosperms (through putative magnolid
pollen), the Barremian-Aptian boundary (125 Ma) appearance of eudicots (based on their
unique triaperturate pollen), and the appearances of hamamelids and rosids in the Albian-
early Cenomanian (97–112 Ma) as an orderly sequence, and a pattern that any claim of a
pre-Cretaceous angiosperm origin must confront.
During the last five years, we have seen additional progress on angiosperm
phylogenetics, and some of the patterns emerging prove more difficult to fit into the
coherent sequence described by Crane et al. (1995). In the palaeobotanical community,
increasing morphological diversity in seeds, pollen, and fruits has been documented from
comparatively old geological deposits (Friis et al. 1999, 2000, 2001). A second
development is that derived angiosperm lineages are being documented from increasingly
older geological deposits. Crepet and Nixon (1998), for example, documented Clusiaceae
from Turonian (88–90 Ma) deposits of New Jersey, Keller et al. (1996) and Herendeen et
al. (1999) documented Actinidiaceae from Campanian (74–83 Ma) and Santonian (83–87
Ma) deposits. Herendeen et al. (1999) suggested a possible affinity to Apiaceae/Araliaceae
for one of their Santonian fossils, and Basinger and Dilcher (1984) documented a possible
Rosaceae/Rhamnaceae from the early Cenomanian (94–97 Ma) of Nebraska.
In parallel to this development, the neobotanical community has seen an explosion in
the amount of molecular data addressing the relationships of extant lineages, and the
hierarchical framework has been transformed from a mere skeleton into a more rigorous
one with most extant families represented, and relationships receiving an increasing
amount of support (Soltis et al. 1997, 1999, 2000; Qiu et al. 1999; Chase et al. 2000; Qiu
2000; Savolainen et al. 2000a,b). The full impact of the palaeobotanical development can
only be appreciated by considering the patterns emerging from these molecular analyses,
and Figure 8.1 illustrates this by indicating the putative positions of some key fossil taxa
on a phylogram representing one of the more than 8000 most parsimonious trees resulting
from the analyses by Soltis et al. (1999). Groups recently documented from Cenomanian-
Campanian deposits are highly derived and nested well inside rosids and asterids, and the
comparatively high levels of sequence divergence separating the Cenomanian-Campanian
fossils (taxa 5–8; 74–97 Ma) and the fossil taxa (taxa 1–4) constituting the orderly
sequence of Crane et al. (1995) indicate one of three possible conclusions. Either the
putatively rapid and explosive diversification of early angiosperms was accompanied by
much more rapid molecular change than subsequently has occurred, or cladogenesis in
early angiosperms took place earlier than our current fossil-based estimates indicate, or
the assignment of these fossils to derived angiosperm families is incorrect.
By accepting this argument, that the levels of sequence divergence on our phylograms
indicate problems, we rely in one way or other on embracing the concept of molecular
clocks (Zuckerhandl and Pauling 1962, 1965). However, all methods, whether they use
rigorous or more relaxed clock assumptions for inferring time from divergence in DNA
sequence data, suffer from their own problems. Some are associated with small datasets
ANGIOSPERM DIVERGENCE 147