Magnoliid (or "ranalean") families (comprising the subclass Magnoliidae in the sense of Cronquist, or the subclasses Magnoliidae plus Ranunculidae in the sense of Takhtajan) have received foremost attention by botanists because they are believed to represent the most primitive group of extant angiosperms. Since the time of Arber and Parkin (1907), the flower of Magnolia has been viewed as the archetype of an angiosperm flower. The work of Bailey and his coworkers on the morphology and relationships of the ranalean families has revealed highly interesting traits, several of which were considered archaic, such as the incompletely closed carpels of Degeneria, the flat, leaflike stamens of the Magnoliaceae and related families, and the vessel-less wood of the Winteraceae, Trochodendraceae, and other families and genera. More recently, the work of Endress and others has permitted the elaboration and partial correction of these views. Friis and Endress (1990) have presented a comprehensive picture of the evolution of magnoliid floral structures, which incorporates the growing body of information from fossils.
Among magnoliid families a major break exists between the predominantly woody "lower magnoliids", characterized by ethereal oil cells and monosulcate (-derived) pollen, by and large corresponding to Takhtajan's (1987) subclass Magnoliidae, and the predominantly herbaceous "higher magnoliids", which lack oil cells, have tricolpate pollen and correspond to Takhtajan's (1987) subclass Ranunculidae. Both groups are held together by the possession of benzylisoquinoline alkaloids, which in the lower magnoliids are accompanied, or can be substituted for, by the biogenetically closely related neolignans (Gottlieb et al. 1989). In the higher magnoliids the latter are negligible, whereas benzylisoquinoline alkaloids are significantly diversified. The position of the TVochodendrales is problematic (see p. 7).
Magnoliales. The "core group" corresponds to the orders Magnoliales anchor Annonales of most authors. Relatively large floral structures with numerous spirally arranged elements are characteristic for this group; often they are reduced to form trimerous whorls. The latter condition is sometimes seen as preceding the former (Dahlgren 1983; Do-
noghue and Doyle 1989), while Kubitzki (1987b) argued for the reverse sequence. Space restrictions on very small or zygomorphic floral apices lead to a more or less unordered phyllotaxy in the androecia of some Winteraceae and Annonaceae. This appears to be a reductive development, since both families include representatives with spirally arranged androecia. The flowers of various families of the Magnoliales (Degeneriaceae, Himantandraceae, Eupo-matiaceae, Austrobaileyaceae) possess highly elaborate inner staminodes that play an important role in the pollination process (Endress 1986 b). They are capable of movements, like the perianth of Magnolia. Floral movements may require organs with broad bases, so that the peculiar shape of the staminodes and stamens, which because of their leaf-like appearance formerly were interpreted as primitive, may have a functional basis. Likewise, the high number of floral organs in the Magnoliales may be related to their pollination syndrome, mainly cantharophily. In these respects the flowers of the Nymphaeaceae are similar.
Other important traits of the Magnoliales include the frequent occurrence of P-type sieve element plas-tids (possibly apomorphic: Behnke 1988), tri- or multilacunar nodes, pluricarpellate apocarpous gy-noecia (excepting Degeneria and Myristicaceae), several ovules per carpel (Himantandra and Myristicaceae excepted), ruminate seeds, and median prophylls. The peculiar bilobed carpel apices found in the Annonaceae, Austrobaileya, and the mid-Cretaceous Lesqueria may be an ancient trait.
Although somewhat more isolated, both the Annonaceae and Myristicaceae are best placed in the Magnoliales. These families are often thought to be closely related but exhibit a high degree of distinctness, the Myristicaceae being strongly heterobath-mic. The position of the Canellaceae remains problematic: is their monadelphous androecium homologous to that of the Myristicaceae, or do their exotestal seeds indicate a relationship to the Winteraceae (with which the Canellaceae share sesquiterpenes of the drimane type) and to the Illiciales?
Donoghue and Doyle (1989) argue that the Magnoliales are held together, among other characters, by their granular exine structure, and that the columel-late exine of the remaining dicotyledons is an advanced character. Since both granular and columel-late exines occur in the Annonaceae, Myristicaceae, and Canellaceae, the transition from the granular to the columellate state could then have occurred several times.
Laurales. An order Laurales is often based on the combination of uniovulate carpels, unilacunar nodes, and opposite leaves (the spiral phyllotaxy of most
Lauraceae would then be a reversal). The Calycan-thaceae are certainly the most aberrant element, yet not as primitive as claimed by Loconte and Stevenson (1991), but rather autapomorphic. Austrobai-leya is problematic, but magnolialean rather than lauralean (if one wishes to make that distinction), though hardly the "first branch" of the Magnoliales, as suggested by Loconte and Stevenson (1991); except for the possession of inner staminodes, which unite Austrobaileya and the Calycanthaceae (Donoghue and Doyle 1989), the floral structure of the two families has little in common. Although the Chloranthaceae are often placed in the Piperales, they seem to be lauralean. Endress (1987 a) found that they share the greatest overall similarity with the Trimeniaceae, then with Amborella, then with other families of the Laurales, and least with the Piperales.
The association of the Monimiaceae, Lauraceae, Gomortega, and Hernandiaceae as "higher" Laurales is generally acknowledged; they are held together, i. a., by filament-bearing stamens with lateral appendages and inaperturate pollen. It can be questioned whether the Laurales are sufficiently distinct from the Magnoliales in terms of innovations to merit ordinal rank; most of their characters are reductive (small, whorled flowers, uniovulate carpels, unilacu-nar nodes, inaperturate pollen) and may have been acquired along somewhat different lines. In the light of secondary metabolism a clear distinction between the two orders is impossible (Gottlieb et al. 1989, and p. 22), and an inclusive order Magnoliales would eliminate the difficulties arising from borderline cases. Moreover, the formal separation of the Laurales from the Magnoliales would make the latter paraphyletic.
Winteraceae/Illiciales. Striking similarities between Drimys and Illicium (and the Canellaceae) were first noted by Miers (1858), and a close relationship between the Winteraceae and the iniciales was suggested by Thome (1974) and accepted by Takhtajan (1980a). Donoghue and Doyle (1989) link the two groups on account of their reticulate exine sculpture with tall muri and short columellae, branched foliar sclereids, exotestal seeds, and possibly their chromosome number (x = 13 or 14). While ethereal oil cells are present in both of them, benzylisoquinoline alkaloids are missing, and neolignans, which often co-occur with or substitute them and indicate the presence of the same biogenetic pathway (Gottlieb et a. 1989), are present only in the Illiciales. Still, both the Winteraceae and the Illiciales may safely be considered genuine magnoliid taxa. In the Winteraceae the pollen is unisulcate-derived ulcerate. The Illiciales received considerable interest because of their peculiar tri(syn)colpate pollen. Various authors have suggested that the Illiciales were a possible link between the unisulcate(-derived) magnoliids and the tricolpate(-derived) Ranunculales, Centrospermae, or Polygonaceae.
Recently, the idea of a relationship between the Winteraceae and Illiciales has received support through the study of fossil pollen of winteraceous affinity from the early Cretaceous by Doyle et al. (1990a,b). The possibility of deriving tricolpates from Illiciales-type pollen, on the other hand, was discounted. The proposed sister-group relationship between Illiciales and Winteraceae implies that the monads of the former are secondarily derived from tetrads. This would elegantly explain puzzling features of Illiciales pollen (Fig. 127, p. 590). The fact that the exine thins towards the "blank" (proximal) pole (where colpi do not fuse) can be interpreted as a vestige of the tetrad condition. The three fused colpi are arranged according to Garside's Rule (instead of Fischer's Rule, according to which they are formed in most other tricolpates), and possess unusual ridges. It is likely that this peculiar aperture arrangement originated from a trichotomous condition, in which the extensions of the triradiating aperture shifted into an equatorial position. Formerly, this mechanism was envisaged as a possible mode of origin of "normal" tricolpates, but now it seems to be a rather exceptional case best known from triaperturate palm pollen. At present, it is generally acknowledged that "normal" tricolpates originated from unisulcates through abrupt changes of the symmetry of the pollen grains in the tetrad (see p. 7). While the idea of a close relationship between Winteraceae and Illiciales thus gains momentum, the pollen apertures of the latter become irrelevant for the derivation of tricolpate dicotyledon pollen. Another corollary of the work of Doyle et al. (1990 b) is that it indirectly supports the older hypothesis of a secondary loss of vessels in vessel-less dicotyledons, a hypothesis that, according to Carlquist (1983), appears unlikely on wood anatomical grounds.
Piperales. After the removal of the Chloranthaceae, this order includes the two closely related families Piperaceae and Saururaceae, held together by ortho-tropous ovules, sclerotic tegmen, minute pollen, and perisperm replacing most of the endosperm. It has often been argued that the Lactoridaceae may be close to the Piperales; their stipules and swollen nodes (but not the two trace-unilacunar structure of the latter) would be indicative of such a position. Interestingly, Carlquist (1990) has now shown that the wood of Lactoris is very similar to that of the Piperales. However, the absence of perisperm and various other piperalean traits of Lactoris would be more compatible with its (somewhat questionable) allocation to the Magnoliales s.l.
Dahlgren (1983) suggested a close relationship between the Piperales and Nymphaeales, which he considered descendents of the "basically trimerous pre-Magnoliiflorae". This suggestion was based on the common possession of perisperm, whose independent origin in the two orders was considered unlikely. I cannot follow Dahlgren for the reason that the Piperales are genuinely magnoliidaean (oil cells, unisulcate-derived pollen, neolignans in both families, benzylisoquinoline in Piperaceae), while the Nymphaeales are not; they agree with magnoliid families mainly in plesiomorphic characters.
4 But hardly secondarily woody, as implied by the "paleoherb concept" (see p. 9): although subfam. Asaroideae, which has the most primitive flowers, is fully herbaceous, tribe Bragantieae, equally conservative in its floral structure, is more or less woody but never fully herbaceous, and tbc same is true of the conservative genera Isotrema and Pararistolochia.
5 The presence of the ellagitannin isoterbechin in the mediterranean Cytinus hypocystis, characterized by Schildknecht et al. (1985), strongly militates against this alignment, and suggestions of a santalalean (Thorne 1983) or ericalean (Goldberg 1986) relationship gain momentum. However, a point of uncertainty is the widespread occurrence of ellagitannins in the genus Cislus, the host of Cytinus, although I consider the possibility of an uptake of these substances from the host as highly unlikely.
atactostelic, but it has been demonstrated that it does not really follow the monocotyledonous pattern. The large, polymerous floral structures with broad stamens are specialized for beetle pollination, although basically the flower construction is archaic. Also the monosulcate(-derived) pollen grains are very primitive, although the atectate condition is hardly ancestral (Ward et al. 1989). The seed coat anatomy is derived and resembles that of the Illi-ciales. The Nymphaeales have a fossil record dating back into the Lower Cretaceous. Because of the purported possession of only one cotyledon, they have often been considered as an extant link between dicotyledons and monocotyledons. Tillich (1990) has now shown that the seedlings of the Nymphaeaceae (several genera studied) are clearly dicotyledonous. All systematists treat the Nymphaeales as part of the ranalean complex, with the notable exception of Walker (1976a), who has elevated them to the rank of a subclass. I cannot find convincing evidence for linking the Nymphaeales to the Magnoliidae. Morphological characters, which appear plesiomorphic (spiral flowers, monosulcate pollen) or specialized for beetle pollination, tell little about relationships. Chemically the Nymphaeales differ fundamentally from all ranalean families in the absence of benzylisoquinoline alkaloids and the presence of pseudoal-kaloids based on sesquiterpenes and of ellagitannins. Ellagitannins are characteristically absent from the magnolialean complex, whereas they are frequent constituents of rosid and dilleniid dicotyledons. Yet in Nuphar, hydrolyzable tannins have been found to possess only a-D-glucopyranose cores, while elsewhere in dicotyledons they are formed by oxidative carbon-to-carbon or carbon-to-oxygen couplings (Ishimatsu et al. 1989). This is certainly an indication that ellagitannins in the Nymphaeales have originated independently from rosid/dilleniid orders. The Ceratophyllaceae, up to the present usually included in the Nymphaeales, have no characters that permit linking them to any other family or order of dicotyledons. Unlike the Nymphaeales, they have no perisperm, and they lack gallotannins and sesquiter-pene-derived pseudoalkaloids, but it is not clear whether these compounds were lost or never present. (In the possible replacement of perisperm by a storage embryo, the Ceratophyllaceae would be paralleled by some barrel cacti.) The polymerous spiral androecium of the Ceratophyllaceae seems to be plesiomorphic, and their fossil record indicates an old age (see Les, p. 249). Our present knowledge of the family would suggest a highly isolated position, although a formal proposal to this end would appear premature.
The chemical evidence for a nonranalean position of the Nymphaeales implies the exclusion of the Ne-
lumbonaceae, because this family contains benzyliso-quinoline alkaloids of a rather primitive type characteristic of the woody magnoliids. The possession of nuclear endosperm development also points to the magnoliids rather than ranunculids, where it is restricted to the Lardizabalaceae; however, the tricol-pate pollen of Nelumbo is the characteristic ranuncu-lalean advance.
Higher Magnoliids. The major advancement of the higher magnoliids is the tricolpate pollen. Among the numerous hypotheses that have been offered for its origin, those considering trichotomosulcate or in-aperturate intermediates have now become very improbable. The most likely explanation for the transition to the tricolpate state is a change in symmetry of the grains in the pollen tetrad (simultaneous type of meiosis), as suggested by Huynh (1976). This would imply the possibility of a multiple origin of tricol-pates. Indeed, hamamelidaean tricolpates may have existed before ranunculidaean tricolpates originated in the magnoliid line (see p. 10). This is in contrast to current phylogenetic models, in which the lower hamamelids are linked with ranunculids (see, e.g. Donoghue and Doyle 1989) or Chloranthaceae (En-dress 1987 a).
Rannnculales (incl. Papaverales). It has been suggested that the entire group has a herbaceous ancestry and includes secondarily woody elements (Cronquist 1988), but this needs to be further substantiated. If the ligneous families are considered primarily woody, they had to come first. Apart from being woody, the Lardizabalaceae are often considered the most primitive family of this order. The reason is that their seed coat structure is similar to that of Illicium, and that they have cellular endosperm development, which they share with lower magnoliids and Nelumbo; all other Ranunculales have nuclear endosperm. The Berberidaceae, Ranunculaceae, and Menispermaceae are closely related to each other, although the latter family shows many structural and chemical advances. Still more advanced are the Papaveraceae, which have evolved laticifers and highly elaborated benzylisoquinoline alkaloids. The Fumariaceae have undergone a specialization leading to highly elaborate floral structures. Lidén (p. 557) insists that Pteridophyllum needs family status; it seems closer to him to the Berberidaceae than to the Papaveraceae. Suggestions for cladistic relationships within the Ranunculales are still contradictory (see Loconte and Estes 1989; Lo-conte and Stevenson 1991).
Was this article helpful?