It is now generally accepted that the floral structure in the catkin-bearing "Amentiferae" is secondarily simplified in adaptation to anemophily. The development of a concept of a woody, predominantly anemo-philous clade6 has led to the exclusion of various alien elements and the inclusion of the (partly) insect-pollinated, nectariferous Trochodendrales and Hama-melidales. It is often thought that these remodeled Hamamelididae form a largely coherent group but this is not universally accepted, as follows from the utterly disparate schemes of relationships of Barabé etal. (1987); Takhtajan (1987); Cronquist (1988); Thome (1989); Dahlgren (1980 a,b), and Wolfe (1989).
The "lower" Hamamelididae, comprising the Trochodendrales and Hamamelidales, share (slightly different forms of) valvate anther dehiscence (Endress 1989 a), a trait already known from the Magnoliales s.l. (Endress and Hufford 1989). According to En-dress (1986b), the three families Trochodendraceae, Eupteleaceae, and Cercidiphyllaceae form a coherent assemblage, the Trochodendrales, to which tentatively the Myrothamnaceae can be added (see Endress 1989 b; Hufford and Crane 1989). This alliance had a rich representation during the Late Cretaceous and Early Tertiary (Friis and Crane 1989), but now it appears to be on the verge of extinction. The group is most remarkable for sharing similarities both with magnoliid and "higher" hamamelid orders, and therefore it is often considered to represent a "transitional" taxon. Vessel-less wood in Trochodendron, "oil" cells in all floral organs of Tetracentron, reminiscent of the ethereal oil cells of the lower magnoliids (and possibly substituted in Trochodendron by scarified, branched idioblasts), chloranthoid leaf teeth, and adaxial prophylls (Cercidiphyllum) point to the lower magnoliids or, at least, to a low level of specialization. In contrast, tricolpate (or derived polyforate) pollen, and presence of ellagic acid are markers of a higher level of dicotyledon organization. The perianth is reduced in the Trochodendrales, but rudimen-tarily present, and it is likely that they evolved from an ancestor, in which it was not yet differentiated into sepals and petals, like in the Magnoliales. The Hamamelidales, in contrast, probably originated from a group that possessed a perianth well-differentiated into sepals and petals, which means that these organs were lost secondarily in many taxa (Endress 1986 a). This would imply that the Trochodendrales were not ancestral to the Hamamelidales; both orders may (or
6 See three symposia: Brittonia 25: 315^105 (1973); Ann. Mo. Bot. Gard. 73: 225-441 (1986), and Crane and Blackmore (1989).
may not) have a common ancestry, although the former has retained more magnoliidaean characters than the latter. This demonstrates how problematic the inclusion of Trochodendrales in the "Hamame-lididae" is; in fact, at one time they formed part of the "Polycarpicae".
Traditionally, the Hamamelidales comprise the apocarpous Platanaceae and the syncarpous Hama-melidaceae. An abundant fossil record of the Platanaceae dates back to the Lower Cretaceous and permits tracing changes in their pollination and dispersal syndrome (Friis and Crane 1989). The Hamamelida-ceae appear in the Upper Cretaceous. The close relationship between these families, which had been challenged by Wolfe (1989), has now been confirmed by Schwarzwalder and Dilcher (1991).
The "higher" Hamamelididae comprise the Fagales (incl. Betulales), Juglandales (incl. Myricales), and probably also the Casuarinales. The Fagales and Juglandales are held together by the possession of strictly dichasial inflorescences, and the accumulation of large quantities of galloyl esters, polyphenols, and triterpenes, as well as other traits. The inclusion of the pinnately leaved Juglandales with their resiniferous ducts and glands has always been a problem, and Thorne (1974) suggested a rosid-ruta-lean affinity for them. However, as Cronquist (1981) has pointed out, the Fagales and Juglandales are so closely linked that they cannot be separated from each other. The pollen morphological divergence in the higher Hamamelididae poses another problem: while pollen grains in most Fagaceae (except Nothofagus) are prolate-tricolp(or)ate (as in the lower Hamamelididae), the Betulaceae, Juglandales (incl. Myricaceae), and Casuarinaceae have oblate, ± tri-porate pollen strongly reminiscent of the Norma-polles group that was characteristic of the Late Cretaceous of eastern North America and Europe. Rhoiptelea pollen, which has been identified from Late Cretaceous beds (Wolfe 1973), is a link between the two types because it is triangular-oblate and at the same time tricolporate with very short colpi. This could mean that a lineage with tricolpate spheroidal to prolate grains, such as in Hamamelidales and Fagaceae (except Nothofagus), gave rise to forms with breviaxial grains (such as in Betulaceae, Nothofagus, the Juglandales and the Casuarinaceae) - probably along somewhat different evolutionary lines (see Doyle 1969; Wolfe 1973, 1989; Batten 1989). However, the hypothesis of considering Hamamelidales and Fagaceae (minus Nothofagus) as pre-Norma-polles relics is disputable for two reasons. Firstly, the shift to oblate Normapolles pollen began in the middle of the Cenomanian, long before there was evidence of the Fagaceae. Secondly, it is hard to see how the pre-Normapolles condition could be retained in the wind-pollinated Hamamelidaceae and Fagaceae. This underscores the uncertainty about the relationships between the lower and the higher Hamamelididae, and within the higher Hamamelididae.
Indications of a rosid-dilleniid relationship of the higher Hamamelididae point in the same direction. This is evident from details in leaf structure, particularly the organization of leaf teeth (Wolfe 1973,1989; Hickey and Wolfe 1975). Cunonioid relationships of the Fagaceae have been claimed by Riiffle (1980) who, mainly on the basis of cuticular resemblances, traced back fagalean fossils into the Cretaceous, where they appear to blend with the palmately lobed and palmately compound Debeya-Dewalquea complex. According to this view, these types formed a bridge between the palmate-leaved platanoids and the pinnate fagaceous types. Strong neobotanical evidence in support of a relationship between the predominantly northern hemisphere Hamamelidales/ Fagales/Juglandales and the predominantly southern hemisphere Cunoniales (Cunoniaceae, Brunellia-ceae, Eucryphiaceae) has been presented by Dickison (1989) and Huber (1990). The Casuarinaceae may be an anemophilous derivative of the latter group.
These multiple relationships would be compatible with the idea of a basal connection among the non-magnoliid lineages (Centrospermae and Asteridae excepted), as often has been claimed. Such an early complex may have been basal to palmately and pinnately veined, pinnatifid and compound-leaved lines eventually differentiating into platanoids, ha-mamelids, early rosids, dilleniids, and cornids. Since the first (hamamelid rather than ranunculid) tricol-pates date back to the Barremian/Aptian boundary, the differentiation of this complex could well have taken place in the Lower Cretaceous, or, at the latest, in the Cenomanian, the earliest stage of the Upper Cretaceous, when the divergence of these lineages was enhanced by their switch to anemophily.
A number of small, mostly monotypic families are often, though not consistently, attached to the Hamamelididae but will not be treated in this Volume. Least problematic are the Leitneriaceae, which have been recognized as a catkin-bearing derivative of the rutalean Simaroubaceae (Petersen and Fairbrothers 1983). Simmondsia may be euphorbialean. The remaining families, Eucommiaceae, Buxaceae, Di-dymelaceae, Daphniphyllaceae, and Balanopaceae, all share some hamamelidalean traits, but they also possess strongly advanced characters that remove them from the Hamamelididae and link them to dille-niid or rosid groups. In nearly all of them tannins are absent, and in several iridoids are present. These two traits are unusual in the Hamamelididae. Eucommia with its guttapercha-containing laticifers, synthesis of iridoids, storage of inulin, unitegmic ovules, and Sta-
chyurus-like leaf architecture (Wolfe 1989) is perplexing, and possibly a link with the Cornales. The Buxaceae, linked with Didymeles, and less clearly with Daphyniphyttum and Balanops, may be bordering the dilleniids.
Was this article helpful?