Doyle, J.A. (Section of Evolution and Ecology, University of California, Davis, California 95616, USA). Seed ferns and the origin of angiosperms. J. Torrey Bot. Soc. 133: 169–209. 2006.—If molecular analyses are correct in indicating that Gnetales are related to conifers and no other living gymnosperm group is directly related to angiosperms, studies on the origin of angiosperms must focus on fossil taxa, including “seed ferns.” Some authors have homologized the angiosperm carpel with the cupule of seed ferns, but because angiosperm ovules have two integuments rather than one, cupules are more likely to be homologous with the outer integument. Cupules of the earliest seed ferns may be derived from fertile appendages of “progymnosperms,” but those of later taxa appear to be modified leaves or leaflets, with ovules borne on the abaxial surface in some (peltasperms, corystosperms), the adaxial surface in others (glossopterids, Caytonia). Positional relationships and developmental genetic data suggest that the bitegmic ovule is comparable to a cupule with adaxial ovules. Analysis of a critically revised morphological data set for seed plants indicates that trees in which Gnetales are nested in conifers, as in molecular analyses, are almost as parsimonious as those in which Gnetales are linked with angiosperms, suggesting that the molecular arrangement should be accepted. When living taxa are constrained into the molecular topology, angiosperms are linked with glossopterids, Pentoxylon, Bennettitales, and Caytonia, supporting the homology of the cupule and the bitegmic ovule. Origin of the carpel poses more problems; it could correspond to the leaf portion of the glossopterid leaf-cupule complex, but its homologies in Caytonia are more obscure. New data on currently unknown characters of glossopterids, “Mesozoic seed ferns,” and Bennettitales are needed to test these hypotheses.