Despite much effort that has been devoted to resolving the phylogenetic interrelationships among major groups of angiosperms, this is still a controversial issue. The monocot hypothesis of Burger (1981), based on vegetative similarities between pterido-phytes and monocotyledons, and claiming the primacy of monocotyledons in angiosperm evolution, has found no support at all. The ranalean hypothesis, according to which the magnoliids are the direct ancestor of the monocotyledons and dicotyledons, is almost universally (and uncritically) accepted. Much effort has been devoted to identifying a precise starting point within magnoliid families for the derivation of "higher", i.e., tricolpate dicotyledons. The iniciales, which for a long time had attracted the interest of phylogeneticists because of their apparently tricolpate (i. e., syncolpate) pollen, can now be discounted as a model for the derivation of the tricolpates (see p. 5). However, the idea that the Ranunculales are most closely related to the Magnoliales and related orders is well founded, and the tricolpate ranuncula-lean pollen must have originated in the magnoliid lineage. Since tricolpate angiosperm pollen appears in the fossil record by the Barremian/Aptain boundary (Brenner 1976; Hickey and Doyle 1977), i.e., before the time of a notable diversification of the magnoliids, it is likely that these early tricolpates represented hamamelids, whereas ranunculalean tricolpates may have originated later. In other words, if the hamamelid tricolpates antedated the ranunculalean tricolpates, a multiple origin of tricolpates would have to be postulated.
Another controversial matter is the derivation of hamamelid, dilleniid, and rosid families. According to some authors (including Cronquist 1981,1988, and Takhtajan 1987), these groups (and the Caryophylli-dae) represent parallel branches that evolved from a magnoliid ancestor. This is the essence of the dominating "ranalean" hypothesis, now also supported in terms of enzymatic regulation (see Gottlieb et al., this Vol. p. 30). In a somewhat different vein, Ehrendorfer (1989) argues that the lower Hamamelididae were the major starting point of nonranalean dicot evolution, which took place in two independent phases. In one lineage, the syndrome of anemophily was ela borated, leading to higher Hamamelididae. In a second development, the lower Hamamelididae gave rise to multistaminate zoophilous flowers with double perianths and various structures for nectar production; this radiation led to dilleniid and rosid families. This idea is incompatible with suggestions of authors such as Thorne (1983) and Dahlgren (see Dahlgren 1989 a,b), who more or less completely dismantled the subclass Hamamelididae.
Donoghue and Doyle (1989) tried to resolve these conflicts by a numerical parsimony analysis. Their study raises many interesting points, but also reflects several conventional (though not necessarily acceptable) views (e. g., close relationships between ranun-culids and hamamelids and between Nymphaeales and monocots). Their concept of "paleoherbs", characterized by palmate leaf venation, anomocytic sto-mata, two perianth whorls, and trimerous (or secondarily multiplied) flowers, is most remarkable. It brings together Aristolochiaceae, Piperales, Nymphaeales (!), Lactoris, and the monocotyledons. Fossil evidence (Taylor and Hickey 1990) points to the possibility that such palmately veined, rhizomatous herbs may have belonged to the earliest angiosperms.
A truly alternative view of relationships among major groups of angiosperms is that of Huber (1982, 1990).7 He perceives the major gap in angiosperms not between monocots and dicots, but between a group comprising the magnolialean families in the widest sense (Magnoliidae plus Ranunculidae), the Centro-spermae, and the monocotyledons, and a second group containing the rest of the dicotyledons. Huber speaks of the "First Principal Group" (FPG) and the "Second Principal Group" (SPG), while Kubitzki and Gottlieb (1984 a,b), who have seconded this proposal from the phytochemical point of view, use the terms "Magnolialean" and "Rosiflorean Block". Huber (1982,1990) argues that remnants of gymnospermous characters (angiosperm plesiomorphies) would have to be expected in both the monocotyledons and dicotyledons, if this was the primary division in angiosperms. However, apart from unisulcate(-derived) pollen, plesiomorphic characters are largely absent
7 It is implicit in Hubert concept that multiple relationships between taxa can be recognized and that parallelism is not discarded as mere "evolutionary noise". The abundance of parallelism at all taxonomic levels (see the dismayingly high incidence of parallelism reported by Sanderson and Donoghue 1989) provides support for the idea that evolution proceeds by changes in the expression and reorganisation of the genetic material, i.e., by regulatory mutations, rather than by the appearance of evolutionary novelties (see, e.g., Bachmann 1988, and Kubitzki etal. 1991). Therefore we cannot expect that evolutionary novelties originating in a taxon will invariably form part of the phenotypes of all its descendents; they will rather be realized in some of them and suppressed in others.
from the monocotyledons, while they are scattered throughout the dicotyledons. In addition, several dicotyledonous characters "invade" the monocotyledons, whereas monocotyledonous characters are rare in the dicotyledons and are restricted to the FPG. Monocotyledonous advances, most notably median prophylls, appear frequently in the Magnoliales, Aris-tolochiales, and Piperales. This is understandable, according to Huber (1990), if these orders differentiated before the final branching-off of the monocotyledons from the magnoliids, but after the primary radiation of the dicotyledons. These ideas are not completely new: Already in 1962 Bate-Smith pointed to the notable absence of trihydroxylated phenolic compounds (mainly myricetin, leucodelphinidin and ellagic acid) from the ranalean families, monocotyledons and Centrosper-mae.8 Kubitzki and Gottlieb (1984 a, b) extended this concept by pointing to the prevalence of shikimate-derived compounds in the FPG, best exemplified by phenylalanine-derived alkaloids and amines [benzyl-isoquinolines (QQN + QQN), phenylethylisoqui-no lines (QQN + QQN), Amaryllidaceae alkaloids (Q Q N + Q Q N), betalains, and mescaline (QQN)]. This chemical argument supports the inclusion of the Centrospermae in the FPG as a taxon closely related to the Ranunculidae, with which they share the early transition towards herbosity9 and tri-colpate pollen tending to evolve into polycolpate-polyforate grains (van Campo 1976).
Although the FPG comprises only one third of all angiosperm species, it surpasses the SPG in diversity of life styles and structures. Gymnospermous characters are predominantly retained in the FPG, which on the whole appears to be more conservative than the SPG, but primitive wood and seed coat structures are represented in both principal groups. Several allegedly apomorphic traits of the SPG are also present in the FPG, most notably nectar secreting disks (Centrospermae) and tricolpate pollen.
Huber's concept does not imply the abandonment of the formal subdivision of angiosperms into dicotyledons and monocotyledons. The latter can be clearly circumscribed on the basis of their monoco-
8 Trihydroxylation occurring in lignoids (lignans etc.) of the FPG does not invalidate this concept because of the different biosynthetic origin of hydroxylation in flavonoids and lignoids; see Gottlieb (1992).
9 A characteristic attribute of the herbaceous condition of the Centrospermae is the absence, or rarity, of proanthocyanidins and the utilization of cinnamate, normally used in Iignin synthesis, for the production of other lignoids such as coumarins and lignans (Kubitzki 1987 a). The existence of anomalous secondary thickening per se would be a poor marker for secondary woodiness, while such is strongly indicated by the rayless condition uniting Chenopodiaceae and Amaranthaceae (S.Carlquist, pers. commun. 1991).
tyledonous embryo, peculiar sieve-tube plastids, and additional traits. However, the division into FPG and SPG seems to reflect a more basal differentiation of the angiosperms.
Recent findings in the field of paleobotany give support to the "Principal Group" concept rather than to the ranalean hypothesis, according to which all angiosperms originated via magnoliid forms: Friis and Crane (1989) state that "tricolpate pollen diagnostic of the higher dicotyledons occurs very early in the angiosperm radiation (Barremian-Aptian boundary) particularly at low paleolatitudes ... This palynologi-cal evidence indicates that fossils relevant to assessing the relationship between magnoliids and ha-mamelids may be preserved in pre-Albian rocks. The early occurrence of tricolpate pollen also suggests that some hamamelid lineages may have diverged while angiosperm diversity was still low..., and probably prior to any extensive angiosperm radiation at the magnoliid clade." Schwarzwalder and Dilcher (1991), after investigating the relationship between Platanaceae and Hamamelidaceae, conclude "that Hamamelidae are at least as primitive as Magno-liidae". This is exactly what has been claimed by the proponents of the FPG/SPG concept.
In conclusion, it seems that progress in the field of angiosperm classification and evolution will greatly profit from comparisons between different conflicting, or at least complementary, models rather than by further focusing exclusively on a single concept, such as the ranalean hypothesis.
Was this article helpful?