James E. Eckenwalder
THE classification of gymnosperms, like that of the flowering plants, is based primarily on structures of the sporophyte associated with reproduction because the gametophyte phases of seed plants are wholly included within sporophytic reproductive structures. The major groups of living gymnosperms, recognized by different taxonomists at ranks ranging from division through class and order to family, are distinguished by fundamental differences in the organization of their reproductive structures. Plants of three of the four extant top-rank gymnosperm taxa, here treated as divisions Cycadophyta, Coniferophyta, and Gnetophyta, grow naturally in North America. The fourth division, Ginkgophyta, was worldwide in the Mesozoic and Tertiary but now has a single extant species, Ginkgo biloba, the maidenhair tree. It is cultivated in North America and other temperate areas, but it is apparently no longer found in natural stands (S.Y. Hsu 1980).
The indigenous North American orders all possess pollen cones separate from seed cones but differ in whether the cones of one or both sexes are simple or compound. A simple cone consists of spore leaves inserted directly on a main axis. Both the pollen cones and the seed cones of the Cycadophyta, but only the pollen cones of the Coniferophyta, are simple. Compound cones have bracts along the main axis that subtend fertile dwarf shoots. In the Gnetophyta, both the pollen and the seed cones are compound. The seed cones of the Coniferophyta are also compound, but their fertile dwarf shoots have been transformed into seed-bearing cone scales that deceptively resemble spore leaves. The work of the Swedish paleobotanist R.Florin (1951) that elucidates the origin of the conifer cone scale is one of the great detective stories in plant morphology.
Florin also argued that the solitary terminal seeds of the Taxaceae had a separate evolutionary history from the compound seed cones of the conifers, and he advocated excluding Taxaceae from the conifers. Recent research weakens some of the force of his arguments (R.A. Price 1990), and some authors see the terminal ovule as the result of reduction, a process analogous to that operating within the undoubtedly coniferous genus Juniperus (cf. P.B. Tomlinson et al. 1989).
Taxonomic controversies are common at all ranks in gymnosperms, and thus the classification adopted here, a modification of the standard classification of R.K.F. Pilger (1926), is far from universally accepted. Many past disputes have been finally laid to rest, but new ones continue to emerge. Disagreements include the assignment of ranks for accepted taxa, as with the rank of the taxa here treated as divisions, and extend to substantial differences in circumscription of taxa, as with inclusion or exclusion of Taxaceae in the conifers.
Taxonomic disagreements concerning familial and generic relationships in Cycodophyta and Gnetophyta hardly affect our view of North American taxa because we have only one genus of each, Zamia and Ephedra. Although controversies surrounded the separation of these genera from Cycadaceae and Gnetaceae, respectively, most gymnosperm taxonomists now seem comfortable with including them in the segregate families Zamiaceae (Cycadophyta) and Ephedraceae (Gnetophyta). Even so, disputes continue over the number of species in each genus, more so with respect to Zamia (J.E. Eckenwalder 1980b) than Ephedra (L.D. Benson 1943), although some taxa of the latter have been treated either as species or varieties. Here, we favor a broad specific circumscription.
Disagreements exist on the circumscriptions of families, genera, and species in the coniferophyta, in addition to the issue of segregation of Taxaceae. Both Pinaceae and Taxaceae are firmly established families in the literature, but the Cupressaceae, as used here, include genera (Sequoia, Sequoiadendron, and Taxodium) that were treated in the segregate family Taxodiaceae in almost all previous North American floras. In contrast to the other widely accepted conifer families, which differ in both reproductive and vegetative traits, Cupressaceae and Taxodiaceae overlapped in all traits except phyllotaxis. The discovery of the Chinese Metasequoia, an apparent close relative of Sequoia with decussate phyllotaxis, which is characteristic of most cupressoids, eliminated even that difference. A suggestion that these two families be merged under the name Cupressaceae (J.E. Eckenwalder 1976) has gradually gained acceptance (J.A. Hart and R.A. Price 1990) as new lines of evidence (e.g., the immunological comparisons made by R.A. Price and J.M. Lowenstein 1989) and new kinds of analyses (e.g., the cladistic analysis by J.A. Hart 1987) seem to confirm the artificiality of the traditional familial separation.
The arrangement of genera within the combined family is poorly established, and neither the scheme by R.K.F. Pilger (1926) nor that by Li H. L. (1953) is satisfactory. The arrangement adopted here is a compromise for North American genera, pending a thorough analysis of the family worldwide. The taxodioid genera are placed first because they display the greatest proportion of presumably primitive characterstics among the North American genera. In contrast to Pilger's scheme, ours segregates Sequoiadendron from Sequoia, as is now almost universally accepted following J.T. Buchholz (1939). Cupressus most closely resembles the three taxodioid genera among the cupressoids, Chamaecyparis is close to Cupressus, and Thuja and Calocedrus seem to display increasing asymmetry of the cones, while sharing the flattened branchlet sprays of Chamaecyparis.
Following Li H. L. (1953), as corrected by R.Florin (1956), we segregate Calocedrus from Libocedrus (which is thus restricted to Southern Hemisphere species), and this too is generally unquestioned today. Juniperus appears to be an independent derivative of a Cupressus-like ancestor, with cones specialized for animal dispersal. The naturalized Australian genus Callitris stands apart from the native North American cupressoids, and it is placed last here because of its apparent divergence from Sequoia, Sequoiadendron, and Taxodium, whereas Pilger made it first in his sequence of Cupressaceae in the strict sense.
At the species level, most genera of Cupressaceae are relatively unproblematic, each having 1--3 North American species, but species concepts in Taxodium and Cupressus are particularly controversial, and here we are conservative in both cases. Furthermore, we do not accept some recently proposed, but not widely accepted, segregate species in Chamaecyparis and Juniperus.
Generic arrangements in the Pinaceae are also debated. There are two major schemes, those by P.van Tieghem (1891) and by F.Vierhapper (1910). Recent evidence (cf. R.A. Price et al. 1987, on immunological distances; M.P. Frankis 1988, on seed characters) has favored van Tieghem's arrangement (R.A. Price 1989), and his view is adopted here. In this arrangement, Abies and Tsuga are the sole North American representatives of subfamily Abietoideae. A recent suggestion by C.N. Page (1988) to take up the genus Hesperopeuce for Tsuga mertensiana is not adopted here, pending further evidence and discussion. Pseudotsuga, with a north Pacific distribution, and Larix, Picea, and Pinus, all circumboreal, are the North American genera of the Pinoideae. Pseudotsuga and Larix are closely linked by aspects of pollination biology (J.Doyle and M.O'Leary 1935), among other features. Picea and Pinus are relatively isolated from the former and from each other.
Species delimitation in Pinaceae has been relatively stable, perhaps because the economic importance of these plants led to early exploitation and taxonomic study throughout the continent. There have been two major exceptions to this near-universal agreement. In a treatment rejected by every subsequent North American author, F.Flous (1937) recognized 12 species of Douglas-fir where others see only Pseudotsuga menziesii. More recently D.K. Bailey (1970, 1987; Bailey and F.G. Hawksworth 1979) has proposed several revisions in the taxonomy of southwestern Pinus, only some of which are adopted here, the others not yet widely accepted. Although species delimitations within the North American Pinaceae have remained largely uncontroversial, there have been numerous nomenclatural disputes, most of which have been settled by the research leading to the U.S. Forest Service's most recent checklist (E.L. Little Jr. 1979).
Taxaceae are so poorly represented in North America, and the species so distantly related to one another, that no controversy has arisen regarding the circumscriptions of the five taxa recognized by R.K.F. Pilger (1903). We follow most contemporary authors in treating our three representatives of Taxus as endemic species rather than subspecies of a circumboreal T. baccata. The two endemic species of the presumably more primitive Torreya were also so recognized by Pilger.