Hunting the Snark: the flawed search for mythical Jurassic angiosperms.

Several recent palaeobotanical studies claim to have found and described pre-Cretaceous angiosperm macrofossils. With rare exceptions, these papers fail to define a flower, do not acknowledge that fossils require character-based rather than group-based classification, do not explicitly state which morphological features would unambiguously identify a fossil as angiospermous, ignore the modern conceptual framework of phylogeny reconstruction, and infer features in the fossils in question that are interpreted differently by (or even invisible to) other researchers. This unfortunate situation is compounded by the relevant fossils being highly disarticulated two-dimensional compression-impressions lacking anatomical preservation. Given current evidence, all supposed pre-Cretaceous angiosperms are assignable to other major clades among the gymnosperms sensu lato. By any workable morphological definition, flowers are not confined to, and therefore cannot delimit, the angiosperm clade. More precisely defined character states that are potentially diagnostic of angiosperms must by definition originate on the phylogenetic branch that immediately precedes the angiosperm crown group. Although the most reliable candidates for diagnostic characters (triploid endosperm reflecting double fertilization, closed carpel, bitegmic ovule, and phloem companion cells) are rarely preserved and/or difficult to detect unambiguously, similar characters have occasionally been preserved in high-quality permineralized non-angiosperm fossils. The angiosperm radiation documented by Early Cretaceous fossils involves only lineages closely similar to extant taxonomic families, lacks obvious morphological gaps, and (as agreed by both the fossil record and molecular phylogenies) was relatively rapid-all features that suggest a primary radiation. It is unlikely that ancestors of the crown group common ancestor would have fulfilled a character-based definition of (and thereby required expansion of the concept of) an angiosperm; they would instead form a new element of the non-angiosperm members of the 'anthophyte' grade, competing with Caytonia to be viewed as morphologically determined sister group for angiosperms. Conclusions drawn from molecular phylogenetics should not be allowed to routinely constrain palaeobotanical inferences; reciprocal illumination between different categories of data offers greater explanatory power than immediately resorting to Grand Syntheses. The Jurassic angiosperm-essentially a product of molecular phylogenetics-may have become the holy grail of palaeobotany but it appears equally mythical.